OpenAI’s biggest backer buys nearly half a million GPUs this year in race to build artificial intelligence systems

Microsoft acquires twice as many Nvidia AI chips as tech rivals

Consultation suggests opt-out scheme for creatives who don’t want their work used by Google, OpenAI and others

UK proposes letting tech firms use copyrighted work to train AI | Artificial intelligence (AI) | The Guardian

In January 2023, shortly after OpenAI released ChatGPT, Pokimane had a terrifying "eureka" moment mid-stream about the future of AI in her line of work.

OpenAI's Sora Is Generating Videos of Real People, Including This Unintentionally Demonic Version of Pokimane

Subscribe to our daily newsletter to keep in touch with the subjects shaping our future.

OpenAI’s Super-Hyped Sora Goes Absolutely Freakshow If You Ask It to Generate Gymnastics Videos

Google has reportedly asked the FTC to investigate Microsoft's exclusive access to OpenAI technologies on its cloud platforms. The FTC questioned Google on Microsoft's practices as part of a broader investigation.

Google wants FTC to stop Microsoft's cloud partnership with OpenAI - NotebookCheck.net News

According to the report, companies that purchase ChatGPT-maker OpenAI's technology through Microsoft may have to face additional charges if they don't already use Microsoft servers to run their operations.

Google asks FTC to break up Microsoft's cloud deal with OpenAI: Report

Speak’s app uses an artificial intelligence “conversational partner” to help people become fluent in English.

OpenAI-Backed Language Tutor Startup Doubles Value to $1 Billion - Bloomberg

OpenAI

openai

OpenAI has recently announced a new subscription plan called ChatGPT Pro, priced at $200 per month. This plan includes access to the advanced o1 model, which now supports images and has improved skills in mathematics and coding. As someone who has been following these developments closely, here are my thoughts on this new offering. Exciting Updates with the o1 Model I think it's really cool that OpenAI is making progress again. Recently, there were concerns that AI development had hit a plateau because it required much more power for even small breakthroughs. With the new o1 model, these fears are eased. The o1 model's ability to handle images and perform better in math and coding is impressive. It shows that AI is still advancing and becoming more capable. Is ChatGPT Pro Worth the $200 Monthly Fee? The price of $200 per month for ChatGPT Pro seems reasonable to me. There will definitely be people who are willing to pay this without hesitation. However, the big question is whether the service is good enough to keep subscribers. Some online reviews mention that ChatGPT Pro is about 11% better in coding compared to the standard version. But considering it's usually 10 times more expensive than the regular subscription, it might not be the best deal for everyone. For some, a cheaper alternative like Claude 3.5 might be more appealing. I'm waiting to see what the first users have to say before making a final judgment. Benefits of the o1 Pro Mode Access to the o1 pro mode is a major plus for users who work with more complex queries. I believe this will have a positive impact, especially for those in scientific fields like math, physics, chemistry, and medicine. With more powerful reasoning capabilities, the o1 pro mode can help achieve breakthroughs and solve difficult problems more efficiently. o1 Model Now Available to All Paid Users Finally, OpenAI decided to make the o1 model available to all paid users and move it out of the preview stage. I'm really excited about this change and can't wait to try it out. It's great to see OpenAI making advanced models more accessible, allowing more people to benefit from their capabilities. New Features: Voice Module and Better Responses ChatGPT Pro also offers unlimited access to the voice module and improved answers to complex questions. However, I'm not sure how many users will regularly use the unlimited voice feature. It might be useful for some, but it may not be a deciding factor for everyone. Rumors of a $2,000 Monthly Subscription There are also rumors about a $2,000 per month subscription. I can see this being a reality for professional, governmental, and scientific users. It would be something between the regular web version and API access, possibly providing access to AGI (Artificial General Intelligence) that can work on tasks for hours. This higher-tier subscription could be very valuable for those who need advanced AI capabilities for their work. Conclusion OpenAI's ChatGPT Pro offers exciting new features and improved models like o1, but the high price tag may limit its appeal to a broader audience. It will be interesting to see how users respond and whether the benefits justify the cost. For those who need advanced AI for complex tasks, ChatGPT Pro could be a great investment. However, others might prefer sticking with more affordable options until they see more feedback from early adopters. OpenAI has recently announced a new subscription plan called ChatGPT Pro , priced at $200 per month . This plan includes access to the advanced o1 model , which now supports images and has improved skills in mathematics and coding. As someone who has been following these developments closely, here are my thoughts on this new offering. ChatGPT Pro ChatGPT Pro $200 per month o1 model Exciting Updates with the o1 Model Exciting Updates with the o1 Model I think it's really cool that OpenAI is making progress again. Recently, there were concerns that AI development had hit a plateau because it required much more power for even small breakthroughs. With the new o1 model, these fears are eased. The o1 model's ability to handle images and perform better in math and coding is impressive. It shows that AI is still advancing and becoming more capable. Is ChatGPT Pro Worth the $200 Monthly Fee? Is ChatGPT Pro Worth the $200 Monthly Fee? The price of $200 per month for ChatGPT Pro seems reasonable to me. There will definitely be people who are willing to pay this without hesitation. However, the big question is whether the service is good enough to keep subscribers. Some online reviews mention that ChatGPT Pro is about 11% better in coding compared to the standard version. But considering it's usually 10 times more expensive than the regular subscription, it might not be the best deal for everyone. For some, a cheaper alternative like Claude 3.5 might be more appealing. I'm waiting to see what the first users have to say before making a final judgment. 11% better in coding 10 times more expensive Claude 3.5 Benefits of the o1 Pro Mode Benefits of the o1 Pro Mode Access to the o1 pro mode is a major plus for users who work with more complex queries. I believe this will have a positive impact, especially for those in scientific fields like math, physics, chemistry, and medicine. With more powerful reasoning capabilities, the o1 pro mode can help achieve breakthroughs and solve difficult problems more efficiently. o1 pro mode o1 Model Now Available to All Paid Users o1 Model Now Available to All Paid Users Finally, OpenAI decided to make the o1 model available to all paid users and move it out of the preview stage. I'm really excited about this change and can't wait to try it out. It's great to see OpenAI making advanced models more accessible, allowing more people to benefit from their capabilities. New Features: Voice Module and Better Responses New Features: Voice Module and Better Responses ChatGPT Pro also offers unlimited access to the voice module and improved answers to complex questions. However, I'm not sure how many users will regularly use the unlimited voice feature. It might be useful for some, but it may not be a deciding factor for everyone. Rumors of a $2,000 Monthly Subscription Rumors of a $2,000 Monthly Subscription There are also rumors about a $2,000 per month subscription. I can see this being a reality for professional, governmental, and scientific users. It would be something between the regular web version and API access, possibly providing access to AGI (Artificial General Intelligence) that can work on tasks for hours. This higher-tier subscription could be very valuable for those who need advanced AI capabilities for their work. $2,000 per month AGI (Artificial General Intelligence) Conclusion Conclusion OpenAI's ChatGPT Pro offers exciting new features and improved models like o1, but the high price tag may limit its appeal to a broader audience. It will be interesting to see how users respond and whether the benefits justify the cost. For those who need advanced AI for complex tasks, ChatGPT Pro could be a great investment. However, others might prefer sticking with more affordable options until they see more feedback from early adopters.

OpenAI Launches ChatGPT Pro: Is the $200 Monthly Subscription Worth It?

OpenAI ChatGPT Proን ይጀምራል፡ የ$200 ወርሃዊ የደንበኝነት ምዝገባ ዋጋ አለው?

OpenAI ChatGPT Pro-nu işə salır: 200 dollarlıq aylıq abunə buna dəyərmi?

OpenAI пуска ChatGPT Pro: Заслужава ли си месечният абонамент от $200?

OpenAI lancerer ChatGPT Pro: Er det $200 månedlige abonnement værd?

OpenAI lanza ChatGPT Pro: ¿Vale la pena la suscripción mensual de $200?

OpenAI-k ChatGPT Pro abiarazten du: merezi al du 200 $ hileko harpidetzak?

OpenAI lansira ChatGPT Pro: Isplati li se mjesečna pretplata od 200 USD?

OpenAI משיקה את ChatGPT Pro: האם המנוי החודשי של $200 שווה את זה?

OpenAI が ChatGPT Pro をリリース: 月額 200 ドルのサブスクリプションは価値があるか?

OpenAI го лансираше ChatGPT Pro: Дали вреди месечната претплата од 200 долари?

OpenAI د ChatGPT پرو پیل کوي: ایا د $ 200 میاشتنۍ ګډون ارزښت لري؟

OpenAI Inotangisa ChatGPT Pro: Iyo $200 yeMwedzi Kunyoreswa Kwakakodzera Iko?

In a bold move highlighting growing tensions between artists and AI developers, a protest group known as "Sora PR Puppets" has leaked access to OpenAI's unreleased text-to-video model, Sora, on the Hugging Face platform. This act of defiance sheds light on alleged exploitative practices within OpenAI's early access program and offers a rare glimpse into the capabilities of the Sora model. Understanding Sora: OpenAI's Text-to-Video Model Sora is OpenAI's advanced text-to-video model designed to generate high-quality videos from textual descriptions. Building upon the technologies behind DALL·E and GPT models, Sora utilizes a recaptioning technique to produce videos that closely adhere to user prompts. The model is capable of creating complex scenes featuring multiple characters, specific motions, and detailed backgrounds, demonstrating an understanding of physical world dynamics. The Leak and Its Implications The leaked version of Sora allowed users to generate 1080p, 10-second video clips, reportedly at faster render times than previously disclosed. Notably, these clips bore OpenAI's watermark, indicating their origin. The temporary availability of Sora on Hugging Face lasted several hours before OpenAI intervened to shut down access. https://x.com/btibor91/status/1861433835528839632?embedable=true Artists' Grievances: Unpaid Labor and Strict Controls The protest group claims that OpenAI enlisted hundreds of artists for unpaid testing, maintaining strict control over all generated content and its dissemination. In an open letter accompanying the leak, the artists stated: "We received access to Sora with the promise to be early testers, red teamers, and creative partners. However, we believe instead we are being lured into 'art washing' to tell the world that Sora is a useful tool for artists." The Verge They further criticized the program as being more about public relations than genuine creative collaboration. OpenAI's Response and Ongoing Developments In response to the leak, OpenAI emphasized that participation in its research preview was voluntary, with no obligation to provide feedback or use the tool. The company stated: "Sora is still in research preview, and we're working to balance creativity with robust safety measures for broader use." The Verge Reports suggest that OpenAI has been training a new version of Sora to address long render times, with rumored features like in-painting and image generation. Industry Context: A Glimpse into AI Video Generation While competitors race to ship AI video tools, Sora has remained behind closed doors, making this leak a dramatic first look at OpenAI's progress. The capabilities seem impressive but do not leap ahead of the competition, and this leak may expose some underlying tension with the creatives they depend on for testing. https://x.com/slow_developer/status/1861440783867388298?embedable=true Conclusion: A Call for Ethical Collaboration The "Sora PR Puppets" leak underscores the need for ethical collaboration between AI developers and the creative community. As AI-generated content becomes more prevalent, establishing fair and transparent practices is crucial to ensure that technological advancements benefit all stakeholders involved. In a bold move highlighting growing tensions between artists and AI developers, a protest group known as "Sora PR Puppets" has leaked access to OpenAI's unreleased text-to-video model, Sora, on the Hugging Face platform. This act of defiance sheds light on alleged exploitative practices within OpenAI's early access program and offers a rare glimpse into the capabilities of the Sora model. Understanding Sora: OpenAI's Text-to-Video Model Understanding Sora: OpenAI's Text-to-Video Model Sora is OpenAI 's advanced text-to-video model designed to generate high-quality videos from textual descriptions. Building upon the technologies behind DALL·E and GPT models, Sora utilizes a recaptioning technique to produce videos that closely adhere to user prompts. The model is capable of creating complex scenes featuring multiple characters, specific motions, and detailed backgrounds, demonstrating an understanding of physical world dynamics. OpenAI The Leak and Its Implications The Leak and Its Implications The leaked version of Sora allowed users to generate 1080p, 10-second video clips, reportedly at faster render times than previously disclosed. Notably, these clips bore OpenAI's watermark, indicating their origin. The temporary availability of Sora on Hugging Face lasted several hours before OpenAI intervened to shut down access. https://x.com/btibor91/status/1861433835528839632?embedable=true https://x.com/btibor91/status/1861433835528839632?embedable=true Artists' Grievances: Unpaid Labor and Strict Controls Artists' Grievances: Unpaid Labor and Strict Controls The protest group claims that OpenAI enlisted hundreds of artists for unpaid testing, maintaining strict control over all generated content and its dissemination. In an open letter accompanying the leak, the artists stated: "We received access to Sora with the promise to be early testers, red teamers, and creative partners. However, we believe instead we are being lured into 'art washing' to tell the world that Sora is a useful tool for artists." "We received access to Sora with the promise to be early testers, red teamers, and creative partners. However, we believe instead we are being lured into 'art washing' to tell the world that Sora is a useful tool for artists." The Verge The Verge The Verge They further criticized the program as being more about public relations than genuine creative collaboration. OpenAI's Response and Ongoing Developments OpenAI's Response and Ongoing Developments In response to the leak, OpenAI emphasized that participation in its research preview was voluntary, with no obligation to provide feedback or use the tool. The company stated: "Sora is still in research preview, and we're working to balance creativity with robust safety measures for broader use." "Sora is still in research preview, and we're working to balance creativity with robust safety measures for broader use." The Verge The Verge The Verge Reports suggest that OpenAI has been training a new version of Sora to address long render times, with rumored features like in-painting and image generation. Industry Context: A Glimpse into AI Video Generation Industry Context: A Glimpse into AI Video Generation While competitors race to ship AI video tools, Sora has remained behind closed doors, making this leak a dramatic first look at OpenAI's progress. The capabilities seem impressive but do not leap ahead of the competition, and this leak may expose some underlying tension with the creatives they depend on for testing. https://x.com/slow_developer/status/1861440783867388298?embedable=true https://x.com/slow_developer/status/1861440783867388298?embedable=true Conclusion: A Call for Ethical Collaboration Conclusion: A Call for Ethical Collaboration The "Sora PR Puppets" leak underscores the need for ethical collaboration between AI developers and the creative community. As AI-generated content becomes more prevalent, establishing fair and transparent practices is crucial to ensure that technological advancements benefit all stakeholders involved.

Artists Leak OpenAI's Sora Video Model, Exposing Controversial Testing Practices

This guide will teach you how to use the Streamlit st.audio_input widget to record your voice on your device microphone paired with the OpenAI Whisper model to transcribe and/or translate your speech to text in English. You can later download the transcribed content as a text file in the .txt format. Prerequisites Basic Python knowledge
Streamlit
OpenAI API key. Sign up for an account What is Whisper Whisper is a trained open-source neural network that approaches human-level robustness and accuracy in English speech recognition. The OpenAI API provides two endpoints: Transcriptions
Translations What is Streamlit From the official website, Streamlit is a faster way to build and share data apps. It is an open-source Python library that helps you build web applications for sharing analytical results, building complex interactive experiences, and iterating on top of new machine-learning models. Streamlit is a top choice for Python developers because it has built-in and convenient methods, from taking in user inputs like text, numbers, and dates to showing interactive graphs using the most popular and powerful Python graphing libraries. Installing Streamlit To run any Streamlit apps, you must first install Streamlit using the command: pip install streamlit Installing Other Libraries Since we are working with transforming audio to text, we need to store our environment variables securely. pip install openai python-dotenv Creating the Environment Variable Create a new file in the root project directory and name it .env. Paste in your OpenAI API key: .env OPENAI_API_KEY="sk-..." Creating the App In your directory, create this file, streamlit_app.py which will contain all the Python code to transcribe and translate our audio and output the resulting text. To initialize an instance of the OpenAI client, copy-paste this code: streamlit_app.py import os
from dotenv import load_dotenv
from openai import OpenAI

load_dotenv()

api_key = os.getenv('OPENAI_API_KEY')

client = OpenAI() The code block connects and reads our secret key in the .env file, making sure we are authenticated as users. P.S.: Using the OpenAI API is not free as you have to buy some credits to use the service. Transcription with Whisper Let's update the streamlit_app.py with the following: streamlit_app.py ...
import streamlit as st

st.logo(
  "logo.png",
  size="medium",
  link="https://platform.openai.com/docs",
)

st.title("Transcription with Whisper")

audio_value = st.audio_input("record a voice message to transcribe")

if audio_value:
  transcript = client.audio.transcriptions.create(
    model="whisper-1",
    file = audio_value
  )

  transcript_text = transcript.text
  st.write(transcript_text) The transcriptions API will convert our audio using the st.audio_input widget to record our voice. If the recording exists, the model, Whisper is used to create the desired file format for the transcription of the audio and output the text using the st.write() function which takes a string and writes it directly into our web app. To use the exact logo at the top left of the app, download this and save it in your project directory. Now, let's run this app with this command in the terminal: streamlit run streamlit_app.py Downloading the Transcribed Text The ability to download the transcribed message for later would serve for record-keeping when you need it. Streamlit offers an input widget that allows for the display of a download button. Back to the streamlit_app.py file, update the codebase with the following: streamlit_app.py ...

  txt_file = "transcription.txt"

  # Initialize session state for download confirmation
  if "downloaded" not in st.session_state:
        st.session_state.downloaded = False

    # Download button
  if st.download_button(
       label="Download Transcription",
       file_name="transcription.txt",data=transcript_text,
  ):
       st.session_state.downloaded = True

    # Show success message after download
  if st.session_state.downloaded:
        st.success("Transcription file downloaded successfully!") The following occurs in the lines of code above: st.session_state in Streamlit allows you to share variables between reruns, for each user session
The transcript_text variable will contain the content of the transcribed text
The txt_file variable with the assigned value, transcription.txt is the file name of the transcribed text when the file is downloaded.
Within the function of the st.download_button(), the label describes to the user what the button is for. The st.success status element displays a success message when the file is saved as shown: Translation with Whisper The process of creating the translation is similar to creating the transcription. The translation endpoint will translate a foreign language into written text in English from the input of the audio file. Copy and paste this code. streamlit_app.py ...

st.header("Translation with Whisper", divider="gray")

audio_translate = st.audio_input("record a voice message to translate")

if audio_translate:
  translate = client.audio.translations.create(
    model="whisper-1",
    file=audio_translate
  )

  st.write(translate.text) If you want to create a file to save your translated audio file as text, you can do the same with the Download transcription button. The complete source code is in this repository and give this app a try to transcribe and translate your voice to text. Good luck! Conclusion Instead of using pre-recorded audio from the internet as seen in the OpenAI docs, Streamlit offers you the opportunity to use your voice and pair it with the transcription and translation endpoints provided by OpenAI to create this outstanding project. The microphone in your device can do so much as technology has made it possible to go beyond using it for communication during meetings and calls. This guide will teach you how to use the Streamlit st.audio_input widget to record your voice on your device microphone paired with the OpenAI Whisper model to transcribe and/or translate your speech to text in English. You can later download the transcribed content as a text file in the .txt format. st.audio_input .txt Prerequisites Basic Python knowledge Streamlit OpenAI API key. Sign up for an account Basic Python knowledge Python Streamlit Streamlit OpenAI API key. Sign up for an account Sign up What is Whisper Whisper is a trained open-source neural network that approaches human-level robustness and accuracy in English speech recognition. The OpenAI API provides two endpoints: Transcriptions Translations Transcriptions Translations What is Streamlit From the official website , Streamlit is a faster way to build and share data apps. It is an open-source Python library that helps you build web applications for sharing analytical results, building complex interactive experiences, and iterating on top of new machine-learning models. official website Streamlit is a top choice for Python developers because it has built-in and convenient methods, from taking in user inputs like text, numbers, and dates to showing interactive graphs using the most popular and powerful Python graphing libraries. Installing Streamlit To run any Streamlit apps, you must first install Streamlit using the command: pip install streamlit pip install streamlit Installing Other Libraries Since we are working with transforming audio to text, we need to store our environment variables securely. pip install openai python-dotenv pip install openai python-dotenv Creating the Environment Variable Create a new file in the root project directory and name it .env . .env Paste in your OpenAI API key : API key .env .env OPENAI_API_KEY="sk-..." OPENAI_API_KEY="sk-..." Creating the App In your directory, create this file, streamlit_app.py which will contain all the Python code to transcribe and translate our audio and output the resulting text. streamlit_app.py To initialize an instance of the OpenAI client, copy-paste this code: streamlit_app.py streamlit_app.py import os
from dotenv import load_dotenv
from openai import OpenAI

load_dotenv()

api_key = os.getenv('OPENAI_API_KEY')

client = OpenAI() import os
from dotenv import load_dotenv
from openai import OpenAI

load_dotenv()

api_key = os.getenv('OPENAI_API_KEY')

client = OpenAI() The code block connects and reads our secret key in the .env file, making sure we are authenticated as users. .env P.S.: Using the OpenAI API is not free as you have to buy some credits to use the service. Transcription with Whisper Let's update the streamlit_app.py with the following: streamlit_app.py streamlit_app.py streamlit_app.py ...
import streamlit as st

st.logo(
  "logo.png",
  size="medium",
  link="https://platform.openai.com/docs",
)

st.title("Transcription with Whisper")

audio_value = st.audio_input("record a voice message to transcribe")

if audio_value:
  transcript = client.audio.transcriptions.create(
    model="whisper-1",
    file = audio_value
  )

  transcript_text = transcript.text
  st.write(transcript_text) ...
import streamlit as st

st.logo(
  "logo.png",
  size="medium",
  link="https://platform.openai.com/docs",
)

st.title("Transcription with Whisper")

audio_value = st.audio_input("record a voice message to transcribe")

if audio_value:
  transcript = client.audio.transcriptions.create(
    model="whisper-1",
    file = audio_value
  )

  transcript_text = transcript.text
  st.write(transcript_text) The transcriptions API will convert our audio using the st.audio_input widget to record our voice. If the recording exists, the model, Whisper is used to create the desired file format for the transcription of the audio and output the text using the st.write() function which takes a string and writes it directly into our web app. transcriptions API st.audio_input transcription of the audio st.write() To use the exact logo at the top left of the app, download this and save it in your project directory. download this Now, let's run this app with this command in the terminal: streamlit run streamlit_app.py streamlit run streamlit_app.py Downloading the Transcribed Text The ability to download the transcribed message for later would serve for record-keeping when you need it. Streamlit offers an input widget that allows for the display of a download button. Back to the streamlit_app.py file, update the codebase with the following: streamlit_app.py streamlit_app.py streamlit_app.py ...

  txt_file = "transcription.txt"

  # Initialize session state for download confirmation
  if "downloaded" not in st.session_state:
        st.session_state.downloaded = False

    # Download button
  if st.download_button(
       label="Download Transcription",
       file_name="transcription.txt",data=transcript_text,
  ):
       st.session_state.downloaded = True

    # Show success message after download
  if st.session_state.downloaded:
        st.success("Transcription file downloaded successfully!") ...

  txt_file = "transcription.txt"

  # Initialize session state for download confirmation
  if "downloaded" not in st.session_state:
        st.session_state.downloaded = False

    # Download button
  if st.download_button(
       label="Download Transcription",
       file_name="transcription.txt",data=transcript_text,
  ):
       st.session_state.downloaded = True

    # Show success message after download
  if st.session_state.downloaded:
        st.success("Transcription file downloaded successfully!") The following occurs in the lines of code above: st.session_state in Streamlit allows you to share variables between reruns, for each user session The transcript_text variable will contain the content of the transcribed text The txt_file variable with the assigned value, transcription.txt is the file name of the transcribed text when the file is downloaded. Within the function of the st.download_button(), the label describes to the user what the button is for. st.session_state in Streamlit allows you to share variables between reruns, for each user session st.session_state The transcript_text variable will contain the content of the transcribed text transcript_text The txt_file variable with the assigned value, transcription.txt is the file name of the transcribed text when the file is downloaded. txt_file transcription.txt Within the function of the st.download_button() , the label describes to the user what the button is for. st.download_button() The st.success status element displays a success message when the file is saved as shown: st.success Translation with Whisper The process of creating the translation is similar to creating the transcription. The translation endpoint will translate a foreign language into written text in English from the input of the audio file. translation endpoint Copy and paste this code. streamlit_app.py streamlit_app.py ...

st.header("Translation with Whisper", divider="gray")

audio_translate = st.audio_input("record a voice message to translate")

if audio_translate:
  translate = client.audio.translations.create(
    model="whisper-1",
    file=audio_translate
  )

  st.write(translate.text) ...

st.header("Translation with Whisper", divider="gray")

audio_translate = st.audio_input("record a voice message to translate")

if audio_translate:
  translate = client.audio.translations.create(
    model="whisper-1",
    file=audio_translate
  )

  st.write(translate.text) If you want to create a file to save your translated audio file as text, you can do the same with the Download transcription button. translated audio Download transcription The complete source code is in this repository and give this app a try to transcribe and translate your voice to text. in this repository this app a try Good luck! Conclusion Instead of using pre-recorded audio from the internet as seen in the OpenAI docs , Streamlit offers you the opportunity to use your voice and pair it with the transcription and translation endpoints provided by OpenAI to create this outstanding project. OpenAI docs The microphone in your device can do so much as technology has made it possible to go beyond using it for communication during meetings and calls.

How to Build a Voice Transcription and Translation App with OpenAI Whisper and Streamlit

The HackerNoon Newsletter: Predicted Outputs: The OpenAI Feature You Probably Missed (11/20/2024)

OpenAI recently introduced a powerful feature called Predicted Outputs, which quietly entered the scene without much attention from the technical media—an oversight that deserves correction. I noticed that they mentioned this on X in their developer account, but it didn't get much publicity. I decided to draw attention to it because it's a really cool and useful feature. Predicted Outputs significantly reduce latency for model responses, especially when much of the output is known ahead of time. This feature is particularly beneficial for applications that involve regenerating text documents or code files with minor modifications. What Are Predicted Outputs? Predicted Outputs allow developers to speed up API responses from Chat Completions when the expected output is largely predictable. By providing a prediction of the expected response using the prediction parameter in Chat Completions, the model can generate the required output more efficiently. This functionality is currently available with the latest gpt-4o and gpt-4o-mini models. How Do Predicted Outputs Work? When you have a response where most of the content is already known, you can supply that expected content as a prediction to the model. The model then uses this prediction to expedite the generation of the response, reducing latency and improving performance. Custom Example: Updating Configuration Files Imagine you have a JSON configuration file that needs a minor update. Here's an example of such a file: {
  "appName": "MyApp",
  "version": "1.0.0",
  "settings": {
    "enableFeatureX": false,
    "maxUsers": 100
  }
} Suppose you want to update "enableFeatureX" to true. Instead of generating the entire file from scratch, you can provide the original file as a prediction and instruct the model to make the necessary change. import OpenAI from "openai";

const config = `
{
  "appName": "MyApp",
  "version": "1.0.0",
  "settings": {
    "enableFeatureX": false,
    "maxUsers": 100
  }
}
`.trim();

const openai = new OpenAI();

const updatePrompt = `
Change "enableFeatureX" to true in the following JSON configuration. Respond only with the updated JSON, without any additional text.
`;

const completion = await openai.chat.completions.create({
  model: "gpt-4o",
  messages: [
    { role: "user", content: updatePrompt },
    { role: "user", content: config }
  ],
  prediction: {
    type: "content",
    content: config
  }
});

// Output the updated configuration
console.log(completion.choices[0].message.content); In this example, the model quickly generates the updated configuration file, leveraging the prediction to minimize response time. Streaming with Predicted Outputs For applications that require streaming responses, Predicted Outputs offer even greater latency reductions. Here's how you can implement the previous example using streaming: import OpenAI from "openai";

const config = `...`; // Original JSON configuration

const openai = new OpenAI();

const updatePrompt = `...`; // Prompt as before

const completion = await openai.chat.completions.create({
  model: "gpt-4o",
  messages: [ /* ... */ ],
  prediction: {
    type: "content",
    content: config
  },
  stream: true
});

for await (const chunk of completion) {
  process.stdout.write(chunk.choices[0]?.delta?.content || "");
} At the time of writing, there are no alternatives to this solution from competitors. OpenAI's Predicted Outputs feature appears to be a unique offering that addresses the specific need for latency reduction when regenerating known content with minor modifications. Tips for Developers The cool thing is that to start using it, you need almost nothing. Just take it and use it by adding just one new parameter to the API request. This makes it very easy for developers to implement this feature in their existing applications. Limitations While Predicted Outputs offer significant advantages, there are important considerations: Model Compatibility: Only the gpt-4o and gpt-4o-mini models support Predicted Outputs. I think that the fact it's only available on these models is not a problem at all, because they are currently the best models from OpenAI, and the original GPT-4 is too slow for this anyway.
Billing Implications: Rejected prediction tokens are billed at the same rate as generated tokens. Monitor the rejected_prediction_tokens in the usage data to manage costs.
Unsupported Parameters:

n (values higher than 1)
logprobs
presence_penalty (values greater than 0)
frequency_penalty (values greater than 0)
max_completion_tokens
tools (function calling is not supported)


Modality Restrictions: Only text modalities are supported; audio inputs and outputs are incompatible with Predicted Outputs. Conclusion OpenAI's Predicted Outputs is a groundbreaking feature that addresses a common challenge in AI applications: reducing latency when the response is largely predictable. By allowing developers to supply expected outputs, it accelerates response times and enhances user experience. In my personal opinion, OpenAI models are not as strong as Anthropic's models. However, OpenAI produces many cool and really necessary solutions in other areas. Features like Predicted Outputs set OpenAI apart from other AI providers, offering unique solutions that meet specific needs in the developer community. OpenAI recently introduced a powerful feature called Predicted Outputs , which quietly entered the scene without much attention from the technical media—an oversight that deserves correction. I noticed that they mentioned this on X in their developer account , but it didn't get much publicity. I decided to draw attention to it because it's a really cool and useful feature. Predicted Outputs Predicted Outputs in their developer account Predicted Outputs significantly reduce latency for model responses, especially when much of the output is known ahead of time. This feature is particularly beneficial for applications that involve regenerating text documents or code files with minor modifications. What Are Predicted Outputs? Predicted Outputs allow developers to speed up API responses from Chat Completions when the expected output is largely predictable. By providing a prediction of the expected response using the prediction parameter in Chat Completions, the model can generate the required output more efficiently. This functionality is currently available with the latest gpt-4o and gpt-4o-mini models. prediction gpt-4o gpt-4o-mini How Do Predicted Outputs Work? When you have a response where most of the content is already known, you can supply that expected content as a prediction to the model. The model then uses this prediction to expedite the generation of the response, reducing latency and improving performance. Custom Example: Updating Configuration Files Imagine you have a JSON configuration file that needs a minor update. Here's an example of such a file: {
  "appName": "MyApp",
  "version": "1.0.0",
  "settings": {
    "enableFeatureX": false,
    "maxUsers": 100
  }
} {
  "appName": "MyApp",
  "version": "1.0.0",
  "settings": {
    "enableFeatureX": false,
    "maxUsers": 100
  }
} Suppose you want to update "enableFeatureX" to true . Instead of generating the entire file from scratch, you can provide the original file as a prediction and instruct the model to make the necessary change. "enableFeatureX" true import OpenAI from "openai";

const config = `
{
  "appName": "MyApp",
  "version": "1.0.0",
  "settings": {
    "enableFeatureX": false,
    "maxUsers": 100
  }
}
`.trim();

const openai = new OpenAI();

const updatePrompt = `
Change "enableFeatureX" to true in the following JSON configuration. Respond only with the updated JSON, without any additional text.
`;

const completion = await openai.chat.completions.create({
  model: "gpt-4o",
  messages: [
    { role: "user", content: updatePrompt },
    { role: "user", content: config }
  ],
  prediction: {
    type: "content",
    content: config
  }
});

// Output the updated configuration
console.log(completion.choices[0].message.content); import OpenAI from "openai";

const config = `
{
  "appName": "MyApp",
  "version": "1.0.0",
  "settings": {
    "enableFeatureX": false,
    "maxUsers": 100
  }
}
`.trim();

const openai = new OpenAI();

const updatePrompt = `
Change "enableFeatureX" to true in the following JSON configuration. Respond only with the updated JSON, without any additional text.
`;

const completion = await openai.chat.completions.create({
  model: "gpt-4o",
  messages: [
    { role: "user", content: updatePrompt },
    { role: "user", content: config }
  ],
  prediction: {
    type: "content",
    content: config
  }
});

// Output the updated configuration
console.log(completion.choices[0].message.content); In this example, the model quickly generates the updated configuration file, leveraging the prediction to minimize response time. Streaming with Predicted Outputs For applications that require streaming responses, Predicted Outputs offer even greater latency reductions. Here's how you can implement the previous example using streaming: import OpenAI from "openai";

const config = `...`; // Original JSON configuration

const openai = new OpenAI();

const updatePrompt = `...`; // Prompt as before

const completion = await openai.chat.completions.create({
  model: "gpt-4o",
  messages: [ /* ... */ ],
  prediction: {
    type: "content",
    content: config
  },
  stream: true
});

for await (const chunk of completion) {
  process.stdout.write(chunk.choices[0]?.delta?.content || "");
} import OpenAI from "openai";

const config = `...`; // Original JSON configuration

const openai = new OpenAI();

const updatePrompt = `...`; // Prompt as before

const completion = await openai.chat.completions.create({
  model: "gpt-4o",
  messages: [ /* ... */ ],
  prediction: {
    type: "content",
    content: config
  },
  stream: true
});

for await (const chunk of completion) {
  process.stdout.write(chunk.choices[0]?.delta?.content || "");
} At the time of writing, there are no alternatives to this solution from competitors. OpenAI's Predicted Outputs feature appears to be a unique offering that addresses the specific need for latency reduction when regenerating known content with minor modifications. Tips for Developers The cool thing is that to start using it, you need almost nothing. Just take it and use it by adding just one new parameter to the API request. This makes it very easy for developers to implement this feature in their existing applications. just one new parameter Limitations While Predicted Outputs offer significant advantages, there are important considerations: Model Compatibility: Only the gpt-4o and gpt-4o-mini models support Predicted Outputs. I think that the fact it's only available on these models is not a problem at all, because they are currently the best models from OpenAI, and the original GPT-4 is too slow for this anyway. Billing Implications: Rejected prediction tokens are billed at the same rate as generated tokens. Monitor the rejected_prediction_tokens in the usage data to manage costs. Unsupported Parameters:

n (values higher than 1)
logprobs
presence_penalty (values greater than 0)
frequency_penalty (values greater than 0)
max_completion_tokens
tools (function calling is not supported) Modality Restrictions: Only text modalities are supported; audio inputs and outputs are incompatible with Predicted Outputs. Model Compatibility : Only the gpt-4o and gpt-4o-mini models support Predicted Outputs. I think that the fact it's only available on these models is not a problem at all, because they are currently the best models from OpenAI, and the original GPT-4 is too slow for this anyway. Model Compatibility gpt-4o gpt-4o-mini Billing Implications : Rejected prediction tokens are billed at the same rate as generated tokens. Monitor the rejected_prediction_tokens in the usage data to manage costs. Billing Implications rejected_prediction_tokens Unsupported Parameters : n (values higher than 1)
logprobs
presence_penalty (values greater than 0)
frequency_penalty (values greater than 0)
max_completion_tokens
tools (function calling is not supported) Unsupported Parameters n (values higher than 1) logprobs presence_penalty (values greater than 0) frequency_penalty (values greater than 0) max_completion_tokens tools (function calling is not supported) n (values higher than 1) n logprobs logprobs presence_penalty (values greater than 0) presence_penalty frequency_penalty (values greater than 0) frequency_penalty max_completion_tokens max_completion_tokens tools (function calling is not supported) tools Modality Restrictions : Only text modalities are supported; audio inputs and outputs are incompatible with Predicted Outputs. Modality Restrictions Conclusion OpenAI's Predicted Outputs is a groundbreaking feature that addresses a common challenge in AI applications: reducing latency when the response is largely predictable. By allowing developers to supply expected outputs, it accelerates response times and enhances user experience. In my personal opinion, OpenAI models are not as strong as Anthropic's models. However, OpenAI produces many cool and really necessary solutions in other areas. Features like Predicted Outputs set OpenAI apart from other AI providers, offering unique solutions that meet specific needs in the developer community.

Predicted Outputs: The OpenAI Feature You Probably Missed

Yatiyat mistuwinaka: OpenAI ukan lurawipax Inas jan uñjkchiti

Прагназаваныя вынікі: функцыя OpenAI, якую вы, верагодна, прапусцілі

Resultados previstos: la función de OpenAI que probablemente no conocía

Predviđeni rezultati: OpenAI značajka koju ste vjerojatno propustili

Sòti Prevwa: Karakteristik OpenAI ou Pwobableman rate

予測出力: おそらく見逃している OpenAI の機能

Болжолдонгон натыйжалар: Сиз өткөрүп жиберген OpenAI функциясы

Willasqa Lluqsiykuna: OpenAI Ruway Yaqapaschá faltarqanki

Zvakafanotaurwa Zvinobuda: Iyo OpenAI Feature Iwe Pamwe Wakapotsa

Rezultatet e parashikuara: Tipari OpenAI që ndoshta ju ka humbur

ዝተገመቱ ውጽኢታት፡ ምናልባት ዝሰኣንካዮ ናይ OpenAI ባህሪ

Bashorat qilingan natijalar: Siz o'tkazib yuborgan OpenAI xususiyati

From OpenAI to Closed AI: Custom Chips Are Closing The Doors—What’s Next? 

Ад OpenAI да закрытага AI: Карыстальніцкія мікрасхемы зачыняюць дзверы - што далей? 

De l'OpenAI a la IA tancada: els xips personalitzats estan tancant les portes: què passa? 

Von OpenAI zu Closed AI: Kundenspezifische Chips schließen die Türen – was kommt als Nächstes?

Από το OpenAI στο Closed AI: Τα προσαρμοσμένα τσιπ κλείνουν τις πόρτες—Τι ακολουθεί; 

De la IA abierta a la IA cerrada: los chips personalizados están cerrando las puertas. ¿Qué sigue?

Az OpenAI-tól a zárt mesterséges intelligenciaig: Az egyedi chipek bezárják az ajtókat – mi a következő lépés? 

Dari OpenAI ke AI Tertutup: Chip Kustom Menutup Pintu—Apa Selanjutnya?

OpenAI からクローズド AI へ: カスタム チップが扉を閉ざす — 次は何が起こるのか?

Kobanda na OpenAI tii na AI oyo ekangami: Ba puces personnalisées ezali kokanga ba portes —What’s Next? 

له OpenAI څخه تړل شوي AI ته: دودیز چپس دروازې بندوي — بیا څه دي؟ 

От открытого ИИ к закрытому ИИ: кастомные чипы закрывают двери — что дальше?

Від відкритого штучного інтелекту до закритого штучного інтелекту: спеціальні мікросхеми закривають двері — що далі? 

Many people — particularly those outside the tech world — were introduced to OpenAI in November 2022, when the company launched its now-ubiquitous ChatGPT tool. However, it has been around for a lot longer than that. OpenAI captured headlines again more recently due to plans for a corporate restructuring that would mean it leaves behind its nonprofit roots. OpenAI’s Journey OpenAI started in 2015 as a nonprofit research organization dedicated to developing artificial intelligence tools that would benefit humanity as a whole, unrestricted by the need to turn a profit. Elon Musk and current OpenAI CEO Sam Altman were involved then, but Musk left in 2018. A year later, OpenAI added a for-profit subsidiary, turning the company into a hybrid for-profit and nonprofit entity in what it called a “capped profit” structure. Leaders at the time believed this change would keep OpenAI true to its roots while allowing it to use market power to grow its operations and revenue. The “capped” part of the structure’s name refers to limitations whereby the profits returned to the company’s investors can only be roughly 100 times what they put into the organization. Now, OpenAI’s structure seems set to change again, this time to a for-profit benefit corporation. This type of business focuses on sustainability and social responsibility in addition to profitmaking. One of the biggest differences about this development is that it would result in OpenAI’s nonprofit board losing control. For now, the nonprofit board has full control over the company’s for-profit arm. Another change is that Altman would receive equity for the first time — which could be worth billions. Moreover, the transition to a for-profit benefit company would remove the cap on investors’ returns. However, an interesting clarification came from an OpenAI representative who said the company’s nonprofit component would still exist and remain a core part of its goals. When will this restructuring occur? That’s anyone’s guess since those involved are still hashing out the details with legal professionals. However, as soon as the news reached major reporting outlets, some people started thinking about what it could mean for OpenAI and artificial intelligence in general. What Will the Restructuring Mean? Some people who have analyzed the matter said OpenAI’s restructuring might mean it will begin operating more like a startup. Investors would probably like that, but could it result in fewer checks and balances affecting products and processes? Contrary to its name, the company has never been upfront about its operations — employees must sign nondisclosure agreements that stop them from raising concerns to federal regulators. Additionally, a group of people working in the AI industry — including former OpenAI employees — signed an open letter explaining that those working on pioneering innovations in artificial intelligence have strong financial incentives to resist oversight and not share knowledge of what their products could or could not do with governments or civil society entities. Numerous employees have recently left OpenAI, with some, speaking anonymously, saying that the catalyst was a lack of trust in Altman and his ability to make AI work safely. Many have kept quiet because of a reported agreement people must sign when leaving the company that prevents them from speaking disparagingly about it. Some former workers who did not sign such paperwork have spoken out about their worries about the businesses’ increasingly powerful AI and the people who want it built not proceeding carefully enough. If OpenAI’s nonprofit board loses its current control, that may make the company’s leaders feel more equipped to proceed with whatever projects garner the most attention. Rushing to get these offerings on the market could also mean it doesn’t take safety as seriously as it should. OpenAI’s Impact on the Future of Artificial Intelligence Now that many AI tools are widely available, more people are insisting on controls that limit their potential applications. Others worry about the future, not necessarily because they believe that artificial intelligence will become smarter than humans, but due to privacy concerns, the prospect of job losses, and the potential for people to misuse AI. The shift to a for-profit model might also raise questions about priorities between profit and ethical AI development. Because of these concerns, transparency becomes even more crucial for the growing number of for-profit AI companies as they try to attract stakeholders — particularly as 60% of people cite transparency as the most important brand trait, according to one study. Embracing transparency can help build trust and reassure stakeholders that OpenAI remains committed to its foundational goals, even as its business model evolves. However, that sentiment could become problematic since OpenAI already has a reputation for being tight-lipped about its activities. People appreciate transparent companies, and it is arguably even more important to show that trait when the business specializes in a technology that the public views as largely foreign and new. For the average person who is merely curious about AI and how to use it, however, they likely won’t feel the impacts of OpenAI’s likely restructuring — at least not at first. Regular consumers will mostly pay attention to which products the company releases next. If those offerings become extremely profitable, such results will signal to its leaders that they should keep doing more of the same. It’s worth highlighting how OpenAI and the industry, in general, are already highly influential, with economic analysts saying generative AI will cause a 7% increase in global gross domestic product, resulting in a difference worth trillions of dollars. A related aspect is how relatively quickly people have gotten to know some of its tools and used them to do business and everyday tasks differently. Many individuals — even those who do not consider themselves tech-savvy — use AI numerous times throughout a typical week. That is almost certainly because of how accessible ChatGPT has become. Its popularity has helped OpenAI come into the mainstream, which arguably has implications for wider society — regardless of the ultimate impact of the eventual restructuring. An Uncertain But AI-Driven Future It’s too early to know the impacts of OpenAI transitioning to a for-profit benefit corporation. However, people can expect some aspects of its internal operations, stakeholder impact, and leadership to change in the coming months and years. We know with more confidence that artificial intelligence will continue playing a role in many people’s lives and shaping the tools they are now accustomed to using. Since OpenAI has enjoyed so much momentum in an increasingly crowded industry, its main task for the foreseeable future is to maintain it. If that happens, it will continue playing a significant role in artificial intelligence’s future, even if people don’t know what that looks like yet. Many people — particularly those outside the tech world — were introduced to OpenAI in November 2022, when the company launched its now-ubiquitous ChatGPT tool. However, it has been around for a lot longer than that. OpenAI captured headlines again more recently due to plans for a corporate restructuring that would mean it leaves behind its nonprofit roots. OpenAI’s Journey OpenAI started in 2015 as a nonprofit research organization dedicated to developing artificial intelligence tools that would benefit humanity as a whole, unrestricted by the need to turn a profit. Elon Musk and current OpenAI CEO Sam Altman were involved then, but Musk left in 2018. A year later, OpenAI added a for-profit subsidiary, turning the company into a hybrid for-profit and nonprofit entity in what it called a “capped profit” structure. Leaders at the time believed this change would keep OpenAI true to its roots while allowing it to use market power to grow its operations and revenue. The “capped” part of the structure’s name refers to limitations whereby the profits returned to the company’s investors can only be roughly 100 times what they put into the organization. roots while allowing it to use roots while allowing it to use Now, OpenAI’s structure seems set to change again, this time to a for-profit benefit corporation. This type of business focuses on sustainability and social responsibility in addition to profitmaking. One of the biggest differences about this development is that it would result in OpenAI’s nonprofit board losing control. For now, the nonprofit board has full control over the company’s for-profit arm. Another change is that Altman would receive equity for the first time — which could be worth billions. Moreover, the transition to a for-profit benefit company would remove the cap on investors’ returns. However, an interesting clarification came from an OpenAI representative who said the company’s nonprofit component would still exist and remain a core part of its goals. nonprofit component would still exist nonprofit component would still exist When will this restructuring occur? That’s anyone’s guess since those involved are still hashing out the details with legal professionals. However, as soon as the news reached major reporting outlets, some people started thinking about what it could mean for OpenAI and artificial intelligence in general. What Will the Restructuring Mean? Some people who have analyzed the matter said OpenAI’s restructuring might mean it will begin operating more like a startup. Investors would probably like that, but could it result in fewer checks and balances affecting products and processes? Contrary to its name, the company has never been upfront about its operations — employees must sign nondisclosure agreements that stop them from raising concerns to federal regulators. must sign nondisclosure agreements must sign nondisclosure agreements Additionally, a group of people working in the AI industry — including former OpenAI employees — signed an open letter explaining that those working on pioneering innovations in artificial intelligence have strong financial incentives to resist oversight and not share knowledge of what their products could or could not do with governments or civil society entities. Numerous employees have recently left OpenAI, with some, speaking anonymously, saying that the catalyst was a lack of trust in Altman and his ability to make AI work safely. Many have kept quiet because of a reported agreement people must sign when leaving the company that prevents them from speaking disparagingly about it. Some former workers who did not sign such paperwork have spoken out about their worries about the businesses’ increasingly powerful AI and the people who want it built not proceeding carefully enough. businesses’ increasingly powerful AI businesses’ increasingly powerful AI If OpenAI’s nonprofit board loses its current control, that may make the company’s leaders feel more equipped to proceed with whatever projects garner the most attention. Rushing to get these offerings on the market could also mean it doesn’t take safety as seriously as it should. OpenAI’s Impact on the Future of Artificial Intelligence Now that many AI tools are widely available, more people are insisting on controls that limit their potential applications. Others worry about the future, not necessarily because they believe that artificial intelligence will become smarter than humans, but due to privacy concerns, the prospect of job losses, and the potential for people to misuse AI. The shift to a for-profit model might also raise questions about priorities between profit and ethical AI development. Because of these concerns, transparency becomes even more crucial for the growing number of for-profit AI companies as they try to attract stakeholders — particularly as 60% of people cite transparency as the most important brand trait, according to one study. Embracing transparency can help build trust and reassure stakeholders that OpenAI remains committed to its foundational goals, even as its business model evolves. 60% of people cite transparency 60% of people cite transparency However, that sentiment could become problematic since OpenAI already has a reputation for being tight-lipped about its activities. People appreciate transparent companies, and it is arguably even more important to show that trait when the business specializes in a technology that the public views as largely foreign and new. For the average person who is merely curious about AI and how to use it, however, they likely won’t feel the impacts of OpenAI’s likely restructuring — at least not at first. Regular consumers will mostly pay attention to which products the company releases next. If those offerings become extremely profitable, such results will signal to its leaders that they should keep doing more of the same. It’s worth highlighting how OpenAI and the industry, in general, are already highly influential, with economic analysts saying generative AI will cause a 7% increase in global gross domestic product, resulting in a difference worth trillions of dollars. A related aspect is how relatively quickly people have gotten to know some of its tools and used them to do business and everyday tasks differently. Many individuals — even those who do not consider themselves tech-savvy — use AI numerous times throughout a typical week. will cause a 7% increase will cause a 7% increase That is almost certainly because of how accessible ChatGPT has become. Its popularity has helped OpenAI come into the mainstream, which arguably has implications for wider society — regardless of the ultimate impact of the eventual restructuring. An Uncertain But AI-Driven Future It’s too early to know the impacts of OpenAI transitioning to a for-profit benefit corporation. However, people can expect some aspects of its internal operations, stakeholder impact, and leadership to change in the coming months and years. We know with more confidence that artificial intelligence will continue playing a role in many people’s lives and shaping the tools they are now accustomed to using. Since OpenAI has enjoyed so much momentum in an increasingly crowded industry, its main task for the foreseeable future is to maintain it. If that happens, it will continue playing a significant role in artificial intelligence’s future, even if people don’t know what that looks like yet.

Inside OpenAI's Path to Becoming a For-Profit Powerhouse

Seatbelts are safety, but what is safety for seatbelts if there are risks they face in certain situations? How far down can safety be designed for systems, with respect to designing safety for safety systems? Technologies, or efficiency systems, do not just advance better than safety can apply, but technologies do not have a basis--naturally--for safety. A human, on average, anywhere on the body would quickly move away from a sudden hot object. Safety is built in, against negative and dangerous experiences, that either automatically or not, it is possible to avoid danger, without having a system for safety in one part of the body, that may then need its own safety system, or have safety [apply] somewhere not everywhere. Simply, experience is the basis of safety for humans as a means of survival in society. Experience also is the source of progress in society as laws are based on the possibility that manypeople will try to avoid the negative experiences that will result from breaking the law. Laws or regulations often build on human affect or experience, such that laws are not made for people who will not be affected--if there is a penalty. So human society is not just intelligence driven, but of experiences--by affect. This is a major contrast with technology. Even with several safety systems in automobiles, they still do not know if they crumple. So while they tryto be safe, it is still for people, not for them, which, assuming it were possible, would have been a better end of safety--so to speak. AI is a technology, which is not one in a physical space--where safety seems easier. Digital is too flexible and of a different phase, that it is possible to make large changes and adjustments without the same efforts in the physical. For example, drawing on a wall and on a PC, or driving simulations and so forth, are different from the physical, in most cases. Social media showed that digital contents can be difficult to regulate, with scale, evasion, and speed. Digital, like other technologies, have no affect, so they experience nothing to indicate that they should not allow themselves to be used for one negative purpose or another. AI, though useful, is already getting misused, with several major stories in the last two years. Making AI safe is a problem preliminary of AI lacking experience, hence affect, before any technical hill. How can AI have affect? How can this affect become the basis for AI alignment, such that whenever it is misused, it can know that there is a penalty for it? How can this be adopted so that it becomes the basis for regulation, rather than the common suggestions of inspection or monitoring? There have been several news articles about departures from OpenAI in the last year, with suggestions about not doing enough for safety and alignment. While those with the experience may be certain of what they are saying, across the industry, there is still no major safety tool that is model-independent, against many of the known risks of AI today. This is not about guardrails or red-teaming, but about the misuses of AI, and not a lot of answers, from any source against those problems. The core of AI safety research could be to look at how human affect works with human intelligence for safety, then to explore how to develop parallels for AI, converting parameters into algorithms, to pave way against this very difficult AI safety problem, mirroring the human mind. There is a recent story on TechCrunch, OpenAI loses another lead safety researcher, Lilian Weng, stating that, "Another one of OpenAI’s lead safety researchers, Lilian Weng, announced on Friday she is departing the startup. Weng served as VP of research and safety since August, and before that, was the head of the OpenAI’s safety systems team. Weng’s departure marks the latest in a long string of AI safety researchers, policy researchers, and other executives who have exited the company in the last year, and several have accused OpenAI of prioritizing commercial products over AI safety." There is a recent feature on Quanta Magazine, Debate May Help AI Models Converge on Truth, stating that, "Let two large models debate the answer to a given question, with a simpler model (or a human) left to recognize the more accurate answer. In theory, the process allows the two agents to poke holes in each other’s arguments until the judge has enough information to discern the truth. Building trustworthy AI systems is part of a larger goal called alignment, which focuses on ensuring that an AI system has the same values and goals as its human users. Today, alignment relies on human feedback — people judging AI. But human feedback may soon be insufficient to ensure the accuracy of a system. In recent years, researchers have increasingly called for new approaches in “scalable oversight,” which is a way to ensure truth even when superhuman systems carry out tasks that humans can’t. Computer scientists have been thinking about scalable oversight for years. Debate emerged as a possible approach in 2018, before LLMs became as large and ubiquitous as they are today." There is a new blogpost by character.ai, Community Safety Updates, stating that, "Moving forward, we will be rolling out a number of new safety and product features that strengthen the security of our platform without compromising the entertaining and engaging experience users have come to expect from Character.AI. These include: Changes to our models for minors (under the age of 18) that are designed to reduce the likelihood of encountering sensitive or suggestive content. Improved detection, response, and intervention related to user inputs that violate our Terms or Community Guidelines. A revised disclaimer on every chat to remind users that the AI is not a real person. Notification when a user has spent an hour-long session on the platform with additional user flexibility in progress." Seatbelts are safety, but what is safety for seatbelts if there are risks they face in certain situations? How far down can safety be designed for systems, with respect to designing safety for safety systems? Technologies, or efficiency systems, do not just advance better than safety can apply, but technologies do not have a basis--naturally--for safety. A human, on average, anywhere on the body would quickly move away from a sudden hot object. Safety is built in, against negative and dangerous experiences, that either automatically or not, it is possible to avoid danger, without having a system for safety in one part of the body, that may then need its own safety system, or have safety [apply] somewhere not everywhere. Simply, experience is the basis of safety for humans as a means of survival in society. Experience also is the source of progress in society as laws are based on the possibility that many people will try to avoid the negative experiences that will result from breaking the law. many Laws or regulations often build on human affect or experience, such that laws are not made for people who will not be affected--if there is a penalty. So human society is not just intelligence driven, but of experiences--by affect. This is a major contrast with technology. Even with several safety systems in automobiles, they still do not know if they crumple. So while they try to be safe, it is still for people, not for them, which, assuming it were possible, would have been a better end of safety--so to speak. try AI is a technology, which is not one in a physical space--where safety seems easier. Digital is too flexible and of a different phase, that it is possible to make large changes and adjustments without the same efforts in the physical. For example, drawing on a wall and on a PC, or driving simulations and so forth, are different from the physical, in most cases. Social media showed that digital contents can be difficult to regulate, with scale, evasion, and speed. Digital, like other technologies, have no affect, so they experience nothing to indicate that they should not allow themselves to be used for one negative purpose or another. AI, though useful, is already getting misused, with several major stories in the last two years. Making AI safe is a problem preliminary of AI lacking experience, hence affect, before any technical hill. How can AI have affect? How can this affect become the basis for AI alignment , such that whenever it is misused, it can know that there is a penalty for it? How can this be adopted so that it becomes the basis for regulation, rather than the common suggestions of inspection or monitoring? How can this affect become the basis for AI alignment There have been several news articles about departures from OpenAI in the last year, with suggestions about not doing enough for safety and alignment. While those with the experience may be certain of what they are saying, across the industry, there is still no major safety tool that is model-independent, against many of the known risks of AI today. This is not about guardrails or red-teaming, but about the misuses of AI, and not a lot of answers, from any source against those problems. The core of AI safety research could be to look at how human affect works with human intelligence for safety, then to explore how to develop parallels for AI, converting parameters into algorithms, to pave way against this very difficult AI safety problem, mirroring the human mind. There is a recent story on TechCrunch , OpenAI loses another lead safety researcher, Lilian Weng , stating that, "Another one of OpenAI’s lead safety researchers, Lilian Weng, announced on Friday she is departing the startup. Weng served as VP of research and safety since August, and before that, was the head of the OpenAI’s safety systems team. Weng’s departure marks the latest in a long string of AI safety researchers, policy researchers, and other executives who have exited the company in the last year, and several have accused OpenAI of prioritizing commercial products over AI safety." TechCrunch OpenAI loses another lead safety researcher, Lilian Weng There is a recent feature on Quanta Magazine , Debate May Help AI Models Converge on Truth , stating that, "Let two large models debate the answer to a given question, with a simpler model (or a human) left to recognize the more accurate answer. In theory, the process allows the two agents to poke holes in each other’s arguments until the judge has enough information to discern the truth. Building trustworthy AI systems is part of a larger goal called alignment, which focuses on ensuring that an AI system has the same values and goals as its human users. Today, alignment relies on human feedback — people judging AI. But human feedback may soon be insufficient to ensure the accuracy of a system. In recent years, researchers have increasingly called for new approaches in “scalable oversight,” which is a way to ensure truth even when superhuman systems carry out tasks that humans can’t. Computer scientists have been thinking about scalable oversight for years. Debate emerged as a possible approach in 2018, before LLMs became as large and ubiquitous as they are today." Quanta Magazine Debate May Help AI Models Converge on Truth There is a new blogpost by character.ai , Community Safety Updates , stating that, "Moving forward, we will be rolling out a number of new safety and product features that strengthen the security of our platform without compromising the entertaining and engaging experience users have come to expect from Character.AI. These include: Changes to our models for minors (under the age of 18) that are designed to reduce the likelihood of encountering sensitive or suggestive content. Improved detection, response, and intervention related to user inputs that violate our Terms or Community Guidelines. A revised disclaimer on every chat to remind users that the AI is not a real person. Notification when a user has spent an hour-long session on the platform with additional user flexibility in progress." character.ai Community Safety Updates

OpenAI Alignment Departures: What Is the AI Safety Problem?

There's been a flurry of news coming out about Microsoft and its relationship with OpenAI ever since the ChatGPT maker raised an additional $6 billion in funding in early October. Publicly, Microsoft has been mum about its exact ownership in OpenAI, but news media, as well as analysts, are now offering a glimpse. But first, the big picture. Sam Altman, who has publicly spoken of the need to raise trillions of dollars, raised $6.6 billion in new funding for his startup at the start of October. There were already reports that OpenAI was looking to raise money, so the company's announcement did not come as a surprise at all. What was surprising, however, was how somber the press release was. At three paragraphs long, OpenAI mentioned how much it raised ($6.6 billion), how much it was worth ($157 billion), and what it needed the money for (mostly corporate fluff about advancing its mission). Inconspicuously missing was a mention of who participated in the funding, and how much, if at all, Microsoft invested in the company. And so of course, the news media was left to fill in the gaps. And fill the gaps it did. According to reports, Thrive Capital put in most of the funds at $1.3 billion and negotiated an option to invest another $1 billion next year if OpenAI meets a certain revenue target. Thrive Capital is led by Josh Kushner, who is the brother of Donald Trump's son-in-law Jared Kushner. Khosla Ventures, Fidelity, and Nvidia were also identified as investors in the funding round. And yes, so was Microsoft, which reportedly invested $750 million in OpenAI. GET PUBLISHED ON HACKERNOON.
Join a global network of  45,000+ published devs, builders,  founders, makers, VCs, hodlrs, and  hackers. Start submitting your tech  stories and tutorials to get  published FREE on HackerNoon — no pop-ups,  no paywalls. Now, three quarters of a billion sounds like an impressive figure until you realize that this is perhaps the least amount of money Microsoft has invested in OpenAI in a single go. Microsoft invested $1 billion in OpenAI in 2019, and then another $10 billion in 2023, with a couple of other capital injections here and there. In total, Microsoft has invested around $14 billion in the creator of ChatGPT, and it's no wonder that Sam Altman has described OpenAI's relationship with the Windows maker as the "best bromance in tech." Well, the best "bromance" in tech is coming to a rude awakening as both sides spar over how much equity Microsoft will receive as OpenAI shifts gears to a for-profit entity. Regardless of how much Microsoft ends up owning, it's certain that it will be less than it used to be, and perhaps, this is by design. The company has never acknowledged how much of OpenAI it controlled, but media outlets pegged that number at 49%, just shy of an outright majority. So when Microsoft announced even more investments in OpenAI, European regulators were considering reviewing whether such financing would trigger scrutiny under the bloc's merger rules. That is now less likely to happen as Microsoft's stake is likely to have been diluted following OpenAI's October financing, bringing two key advantages: less regulatory scrutiny and a hedging of bets should things go south. Microsoft really had no plans to withhold funds from OpenAI, but something changed after the whole leadership fiasco at the company in November 2023. Sam Altman's firing by the company's board raised a lot of eyebrows, and even as Microsoft Satya Nadella facilitated Altman's return to OpenAI, the seed of doubt had already been planted in the CEO's mind. Today, The New York Times reports, relationship between OpenAI and Microsoft are "strained," with questions over the financial viability of OpenAI as it continues seeking both money and computing power to keep operating. So does that mean Microsoft is stepping back? Not quite. Both sides have a symbiotic relationship in which OpenAI depends on Microsoft for cloud computing and Microsoft depends on OpenAI for artificial intelligence. But this interdependence is now causing friction as OpenAI thinks Microsoft charges too much for its service and Microsoft thinks it's too dependent on OpenAI for its technology. "Microsoft could be left behind if it is only using OpenAI technologies,” Gil Luria, an analyst at the investment bank D.A. Davidson, was quoted as saying by the NYT. “It is a real race — and OpenAI may not win it." Meanwhile, analysts at investment bank UBS say Microsoft's relationship with OpenAI may be its most strategic yet as the artificial intelligence startup expects to grow its revenue to $100 billion by 2029. But that growth isn't going to come without its cost, and where will that money go? To Microsoft. Since a vast majority of OpenAI training and inference is run on Azure. So the more OpenAI earnings, the more it spends, and the more Microsoft makes, even if its stake is not as high as it used to be. And even with Sam Altman contracting Oracle for some of his company's computing needs, Microsoft still benefits since it just now needs to spend less on developing the data centers needed to power OpenAI. So if you ever thought that the artificial intelligence story was anything other than an attempt by big companies to empower themselves, here's your wakeup call. It just might be about the money 🤑🤑🤑 Microsoft ranked #4 on HackerNoon's Tech Company Rankings this week. In Other News.. 📰 No, Polymarket Whales Aren't Evidence of Prediction Market Manipulation — via CoinDesk
Notion’s email product is nearing launch — via TechCrunch
Elon Musk says he’ll fix the government under Trump. His track record paints a different picture — via CNN
Nvidia CEO Jensen Huang gets rock star treatment from India's tech enthusiasts — via Reuters
Meta's new facial recognition helps recover accounts, spot fake ads — via Axios
Why Meta and Snap are spending billions on AR glasses— via CNBC And that's a wrap! Don't forget to share this newsletter with your family and friends! See y'all next week. PEACE! ☮️ — Sheharyar Khan, Editor, Business Tech @ HackerNoon *All rankings are current as of Monday. To see how the rankings have changed, please visit HackerNoon's Tech Company Rankings page. Tech, What the Heck!? is a once-weekly newsletter written by HackerNoon editors that combine HackerNoon's proprietary data with news-worthy tech stories from around the internet. Humorous and insightful, the newsletter recaps trending events that are shaping the world of tech. Subscribe here. There's been a flurry of news coming out about Microsoft and its relationship with OpenAI ever since the ChatGPT maker raised an additional $6 billion in funding in early October. Publicly, Microsoft has been mum about its exact ownership in OpenAI, but news media, as well as analysts, are now offering a glimpse. Microsoft Microsoft OpenAI OpenAI But first, the big picture. Sam Altman, who has publicly spoken of the need to raise trillions of dollars, raised $6.6 billion in new funding for his startup at the start of October. There were already reports that OpenAI was looking to raise money, so the company's announcement did not come as a surprise at all. raise raise raised raised already reports already reports What was surprising, however, was how somber the press release was. At three paragraphs long, OpenAI mentioned how much it raised ($6.6 billion), how much it was worth ($157 billion), and what it needed the money for (mostly corporate fluff about advancing its mission). was Inconspicuously missing was a mention of who participated in the funding, and how much, if at all, Microsoft invested in the company. And so of course, the news media was left to fill in the gaps. And fill the gaps it did. According to reports, Thrive Capital put in most of the funds at $1.3 billion and negotiated an option to invest another $1 billion next year if OpenAI meets a certain revenue target. Thrive Capital is led by Josh Kushner, who is the brother of Donald Trump's son-in-law Jared Kushner. most of the funds most of the funds invest invest Khosla Ventures, Fidelity, and Nvidia were also identified as investors in the funding round. And yes, so was Microsoft, which reportedly invested $750 million in OpenAI. Nvidia Nvidia GET PUBLISHED ON HACKERNOON. Join a global network of  45,000+ published devs, builders,  founders, makers, VCs, hodlrs, and  hackers. Start submitting your tech  stories and tutorials to get  published FREE on HackerNoon — no pop-ups,  no paywalls. GET PUBLISHED ON HACKERNOON. GET PUBLISHED ON HACKERNOON . GET PUBLISHED ON HACKERNOON Join a global network of  45,000+ published devs, builders,  founders, makers, VCs, hodlrs, and  hackers. Start submitting your tech  stories and tutorials to get  published FREE on HackerNoon — no pop-ups,  no paywalls. Join a global network of  45,000+ published devs, builders,  founders, makers, VCs, hodlrs, and  hackers. Start submitting your tech  stories and tutorials to get  published FREE on HackerNoon — no pop-ups,  no paywalls. Now, three quarters of a billion sounds like an impressive figure until you realize that this is perhaps the least amount of money Microsoft has invested in OpenAI in a single go. Microsoft invested $1 billion in OpenAI in 2019 , and then another $10 billion in 2023 , with a couple of other capital injections here and there. 2019 2019 2023 2023 In total, Microsoft has invested around $14 billion in the creator of ChatGPT, and it's no wonder that Sam Altman has described OpenAI's relationship with the Windows maker as the "best bromance in tech." Well, the best "bromance" in tech is coming to a rude awakening as both sides spar over how much equity Microsoft will receive as OpenAI shifts gears to a for-profit entity. described described spar spar Regardless of how much Microsoft ends up owning, it's certain that it will be less than it used to be, and perhaps, this is by design. The company has never acknowledged how much of OpenAI it controlled, but media outlets pegged that number at 49%, just shy of an outright majority. So when Microsoft announced even more investments in OpenAI, European regulators were considering reviewing whether such financing would trigger scrutiny under the bloc's merger rules. reviewing reviewing That is now less likely to happen as Microsoft's stake is likely to have been diluted following OpenAI's October financing, bringing two key advantages: less regulatory scrutiny and a hedging of bets should things go south. and Microsoft really had no plans to withhold funds from OpenAI, but something changed after the whole leadership fiasco at the company in November 2023. Sam Altman's firing by the company's board raised a lot of eyebrows, and even as Microsoft Satya Nadella facilitated Altman's return to OpenAI, the seed of doubt had already been planted in the CEO's mind. firing firing Today, The New York Times reports, relationship between OpenAI and Microsoft are " strained ," with questions over the financial viability of OpenAI as it continues seeking both money and computing power to keep operating. strained strained So does that mean Microsoft is stepping back? Not quite. Both sides have a symbiotic relationship in which OpenAI depends on Microsoft for cloud computing and Microsoft depends on OpenAI for artificial intelligence. But this interdependence is now causing friction as OpenAI thinks Microsoft charges too much for its service and Microsoft thinks it's too dependent on OpenAI for its technology. "Microsoft could be left behind if it is only using OpenAI technologies,” Gil Luria, an analyst at the investment bank D.A. Davidson, was quoted as saying by the NYT. “It is a real race — and OpenAI may not win it." Meanwhile, analysts at investment bank UBS say Microsoft's relationship with OpenAI may be its most strategic yet as the artificial intelligence startup expects to grow its revenue to $100 billion by 2029. But that growth isn't going to come without its cost, and where will that money go? To Microsoft. Since a vast majority of OpenAI training and inference is run on Azure. So the more OpenAI earnings, the more it spends, and the more Microsoft makes, even if its stake is not as high as it used to be. And even with Sam Altman contracting Oracle for some of his company's computing needs, Microsoft still benefits since it just now needs to spend less on developing the data centers needed to power OpenAI. contracting contracting Oracle Oracle So if you ever thought that the artificial intelligence story was anything other than an attempt by big companies to empower themselves, here's your wakeup call. It just might be about the money 🤑🤑🤑 empower empower Microsoft ranked #4 on HackerNoon's Tech Company Rankings this week. Tech Company Rankings Tech Company Rankings In Other News.. 📰 In Other News.. 📰 No, Polymarket Whales Aren't Evidence of Prediction Market Manipulation — via CoinDesk Notion’s email product is nearing launch — via TechCrunch Elon Musk says he’ll fix the government under Trump. His track record paints a different picture — via CNN Nvidia CEO Jensen Huang gets rock star treatment from India's tech enthusiasts — via Reuters Meta's new facial recognition helps recover accounts, spot fake ads — via Axios Why Meta and Snap are spending billions on AR glasses— via CNBC No, Polymarket Whales Aren't Evidence of Prediction Market Manipulation — via CoinDesk CoinDesk CoinDesk Notion’s email product is nearing launch — via TechCrunch TechCrunch TechCrunch Elon Musk says he’ll fix the government under Trump. His track record paints a different picture — via CNN CNN CNN Nvidia CEO Jensen Huang gets rock star treatment from India's tech enthusiasts — via Reuters Reuters Reuters Meta's new facial recognition helps recover accounts, spot fake ads — via Axios Axios Axios Why Meta and Snap are spending billions on AR glasses— via CNBC CNBC CNBC And that's a wrap! Don't forget to share this newsletter with your family and friends! See y'all next week. PEACE! ☮️ — Sheharyar Khan , Editor, Business Tech @ HackerNoon Sheharyar Khan Sheharyar Khan *All rankings are current as of Monday. To see how the rankings have changed, please visit HackerNoon's Tech Company Rankings page. *All rankings are current as of Monday. To see how the rankings have changed, please visit HackerNoon's Tech Company Rankings page. *All rankings are current as of Monday. To see how the rankings have changed, please visit HackerNoon's Tech Company Rankings page. Tech Company Rankings Tech Company Rankings Tech, What the Heck!? is a once-weekly newsletter written by HackerNoon editors that combine HackerNoon's proprietary data with news-worthy tech stories from around the internet. Humorous and insightful, the newsletter recaps trending events that are shaping the world of tech. Subscribe here. Tech, What the Heck!? is a once-weekly newsletter written by HackerNoon editors that combine HackerNoon's proprietary data with news-worthy tech stories from around the internet. Humorous and insightful, the newsletter recaps trending events that are shaping the world of tech. Subscribe here. Tech, What the Heck!? is a once-weekly newsletter written by HackerNoon editors that combine HackerNoon's proprietary data with news-worthy tech stories from around the internet. Humorous and insightful, the newsletter recaps trending events that are shaping the world of tech. Subscribe here . here here

Microsoft Probably Doesn't Own That Much of OpenAI Anymore (And That's Ok)

OpenAI — the company behind ChatGPT — held its second annual DevDay in the fall of 2024. It unveiled several massive updates, tools, and features to delight attendees and livestream viewers. What were these changes? Will they benefit developers? Here’s a breakdown of the event and what it means for the future of ChatGPT. OpenAI’s 2024 DevDay DevDay — a portmanteau of developer and day — is a global conference hosted by OpenAI. While 2024’s updates were unveiled in San Francisco on Oct. 1, the company plans to visit London on Oct. 30 and Singapore on Nov. 21. This is in contrast to last year when the only event was in San Francisco. OpenAI’s reasoning for expanding its scope is twofold. For one, it wants to be closer to its global audience. The second reason is that developers requested more time to learn from each other after 2023’s DevDay. So, while there will technically be three events, every major update has already been revealed. The most recent conference was the company’s second. It held its first almost one year prior, in November 2023, where it announced a JSON mode, custom models, and a copyright protection tool, among other things. Since 2023’s DevDay was packed with massive updates, many developers expected 2024 to be just as big, if not bigger. While many expected OpenAI to unveil the next iteration of ChatGPT, which started training in 2024, it announced it would prioritize application programming interface (API) advancements and developer tools. This DevDay focused on ChatGPT-4o and GPT-4o mini, which were released in 2024 as successors to GPT-4. The “o” in GPT-4o stands for omni, emphasizing the model’s multimodality. It outperforms GPT-4 in many ways. It supports more languages,has a 128,000-token context window, and a knowledge cutoff date of late 2023. As artificial intelligence goes, it is among the most powerful, fast models today — and it has just received multiple massive updates. 5 Major Updates Revealed During DevDay Five major updates were unveiled during DevDay 2024. While they became available to developers on Oct. 1, some are only for paying customers. Each has unique implications for developers. 1. Model Distillation Model distillation involves leveraging a larger model to teach a smaller one. OpenAI announced multiple features for this cost-effective training method. The first is stored completions, which lets developers automatically keep input-output pairs on the platform permanently. This enables them to seamlessly generate datasets for distillation. The second model distillation tool is a beta of an Evals product. It lets developers conduct distillation end-to-end on the platform to measure their model’s performance. Instead of manually creating scripts and tinkering with a mess of disparate logging tools, they can easily create and run custom evaluations. 2. Vision Fine-Tuning Vision fine-tuning is a customization feature that makes it possible to fine-tune for GPT-4o and 4o-mini with images — not just text. This way, models will have a stronger understanding of visual input. As a result, they perform better in object recognition, detection, image search, and classification tasks. 3. Prompt Caching OpenAI’s prompt caching provides automatic discounts on input the model has recently seen. Instead of paying full price, developers get 50% off — and faster processing speeds — starting at about 1,024 cached tokens. This feature is available on GPT-4o, GPT4-o mini, o1-preview, and o1 mini. Since 1,000 tokens represent approximately 750 words, companies will save on commonly reused prompts. Notably, specifics on how long the cached storage is saved vary depending on whether the user is active during peak or off-peak hours. However, they don’t have to go out of their way to manually apply the discount. 4. Higher Rate Limit A rate limit controls how often users can access the server to make requests within a certain time frame. OpenAI announced it is doubling its rate limit for GPT-4o, going from 5,000 requests per minute to 10,000. This change may not seem as grand as the others, but it will be just as significant in the long run. 5. Realtime API In the past, developers needed three different platforms to create a speech-to-speech model — one for transcription, one for inference, and one for text-to-speech. With Realtime API, they can now integrate fast, natural speech-to-speech conversations into their applications. They have six preset voices to choose from. Realtime API was previously exclusive to ChatGPT as Advanced Voice Mode. Now, professionals can build models that have lifelike speech-to-speech discussions. The delays are shorter and the voices are clearer. Users can even interrupt the AI midsentence to steer the conversation in a different direction. Applications of an Updated ChatGPT Model distillation is one of the most significant features unveiled during the 2024 DevDay since it allows inexperienced professionals to build robust models. It has applications in health care, finance, and retail. It could even be used in a nontechnical setting like education. Since upwards of 60% of educators already use AI in the classroom, integration would be relatively seamless. Notably, model distillation has proven to be an effective tool for improving open models on various tasks. However, it doesn’t reach the teacher model’s level of performance. This slight drop in accuracy and speed may be unacceptable in high-risk, fast-paced industries. However, users can spend extra time fine-tuning their model if needed. Vision fine-tuning is just as significant as model distillation. Developers could create highly accurate image detection and classification models. Trained models could evaluate X-rays to identify cancerous cells, analyze surveillance video to identify intruders, or improve quality control in high-risk settings like aerospace manufacturing. Prompt caching’s real-world impact isn’t as large, but it is still important. It may even be the feature developers are most excited about. After all, it’s something they’ve been waiting years for. Automatically getting half off commonly used prompts could save individuals and companies a lot of money. Notably, prompt caching isn’t exclusive to GPT-4o and GPT-4o mini — Anthropic’s Claude and Google’s Gemini already have it. For reference, Gemini offers a 75% discount on input tokens. However, the key difference is that it isn’t automatic — users must use CachedContent.create to create a cache and explicitly specify usage when defining the model. Realtime API is last but not least. This feature could revolutionize speech-to-speech conversations, enabling seamless communication between humans and AI. Businesses could leverage it for customer service, scheduling, translation, or summarization. They could use it as an assistant, companion, or support tool. What These Updates Mean for ChatGPT’s Future The significance of focusing on API advancements and developer tools for this year’s DevDay cannot be ignored. Instead of rolling out a flashy new model at the big annual conference to make headlines and generate hype, OpenAI is ensuring its foundation is strong. These updates will encourage external innovation and generate insightful feedback from users. These updates could kick-start a growth period for OpenAI, causing speech-to-speech and image-focused models to trend. At the very least, other AI engineers and businesses will look toward this company as inspiration. Similar updates from competitor models will likely follow in the near future. As for the implications of Realtime API, prompt caching, vision fine-tuning, and model distillation, the one-day conferences in London and Singapore will likely be microcosms of what’s to come. Attendees and livestream viewers should expect to see developer-created content, insights, and questions featured at the upcoming events. OpenAI — the company behind ChatGPT — held its second annual DevDay in the fall of 2024. It unveiled several massive updates, tools, and features to delight attendees and livestream viewers. What were these changes? Will they benefit developers? Here’s a breakdown of the event and what it means for the future of ChatGPT. OpenAI’s 2024 DevDay DevDay — a portmanteau of developer and day — is a global conference hosted by OpenAI. While 2024’s updates were unveiled in San Francisco on Oct. 1, the company plans to visit London on Oct. 30 and Singapore on Nov. 21. This is in contrast to last year when the only event was in San Francisco. plans to visit London on Oct. 30 plans to visit London on Oct. 30 OpenAI’s reasoning for expanding its scope is twofold. For one, it wants to be closer to its global audience. The second reason is that developers requested more time to learn from each other after 2023’s DevDay. So, while there will technically be three events, every major update has already been revealed. wants to be closer wants to be closer The most recent conference was the company’s second. It held its first almost one year prior, in November 2023, where it announced a JSON mode, custom models, and a copyright protection tool, among other things. Since 2023’s DevDay was packed with massive updates, many developers expected 2024 to be just as big, if not bigger. While many expected OpenAI to unveil the next iteration of ChatGPT, which started training in 2024, it announced it would prioritize application programming interface (API) advancements and developer tools. This DevDay focused on ChatGPT-4o and GPT-4o mini, which were released in 2024 as successors to GPT-4. The “o” in GPT-4o stands for omni, emphasizing the model’s multimodality. It outperforms GPT-4 in many ways. It supports more languages, has a 128,000-token context window, and a knowledge cutoff date of late 2023. As artificial intelligence goes, it is among the most powerful, fast models today — and it has just received multiple massive updates. has a 128,000-token context window , has a 128,000-token context window 5 Major Updates Revealed During DevDay Five major updates were unveiled during DevDay 2024. While they became available to developers on Oct. 1, some are only for paying customers. Each has unique implications for developers. 1. Model Distillation Model distillation involves leveraging a larger model to teach a smaller one. OpenAI announced multiple features for this cost-effective training method. The first is stored completions, which lets developers automatically keep input-output pairs on the platform permanently. This enables them to seamlessly generate datasets for distillation. The second model distillation tool is a beta of an Evals product. It lets developers conduct distillation end-to-end on the platform to measure their model’s performance. Instead of manually creating scripts and tinkering with a mess of disparate logging tools, they can easily create and run custom evaluations. 2. Vision Fine-Tuning Vision fine-tuning is a customization feature that makes it possible to fine-tune for GPT-4o and 4o-mini with images — not just text. This way, models will have a stronger understanding of visual input. As a result, they perform better in object recognition, detection, image search, and classification tasks. 3. Prompt Caching OpenAI’s prompt caching provides automatic discounts on input the model has recently seen. Instead of paying full price, developers get 50% off — and faster processing speeds — starting at about 1,024 cached tokens. This feature is available on GPT-4o, GPT4-o mini, o1-preview, and o1 mini. starting at about 1,024 starting at about 1,024 Since 1,000 tokens represent approximately 750 words , companies will save on commonly reused prompts. Notably, specifics on how long the cached storage is saved vary depending on whether the user is active during peak or off-peak hours. However, they don’t have to go out of their way to manually apply the discount. represent approximately 750 words represent approximately 750 words 4. Higher Rate Limit A rate limit controls how often users can access the server to make requests within a certain time frame. OpenAI announced it is doubling its rate limit for GPT-4o, going from 5,000 requests per minute to 10,000. This change may not seem as grand as the others, but it will be just as significant in the long run. it is doubling its rate limit it is doubling its rate limit 5. Realtime API In the past, developers needed three different platforms to create a speech-to-speech model — one for transcription, one for inference, and one for text-to-speech. With Realtime API, they can now integrate fast, natural speech-to-speech conversations into their applications. They have six preset voices to choose from. have six preset voices have six preset voices Realtime API was previously exclusive to ChatGPT as Advanced Voice Mode. Now, professionals can build models that have lifelike speech-to-speech discussions. The delays are shorter and the voices are clearer. Users can even interrupt the AI midsentence to steer the conversation in a different direction. Applications of an Updated ChatGPT Model distillation is one of the most significant features unveiled during the 2024 DevDay since it allows inexperienced professionals to build robust models. It has applications in health care, finance, and retail. It could even be used in a nontechnical setting like education. Since upwards of 60% of educators already use AI in the classroom, integration would be relatively seamless. 60% of educators already use AI 60% of educators already use AI Notably, model distillation has proven to be an effective tool for improving open models on various tasks. However, it doesn’t reach the teacher model’s level of performance. This slight drop in accuracy and speed may be unacceptable in high-risk, fast-paced industries. However, users can spend extra time fine-tuning their model if needed. proven to be an effective tool proven to be an effective tool Vision fine-tuning is just as significant as model distillation. Developers could create highly accurate image detection and classification models. Trained models could evaluate X-rays to identify cancerous cells, analyze surveillance video to identify intruders, or improve quality control in high-risk settings like aerospace manufacturing. Prompt caching’s real-world impact isn’t as large, but it is still important. It may even be the feature developers are most excited about. After all, it’s something they’ve been waiting years for. Automatically getting half off commonly used prompts could save individuals and companies a lot of money. Notably, prompt caching isn’t exclusive to GPT-4o and GPT-4o mini — Anthropic’s Claude and Google’s Gemini already have it. For reference, Gemini offers a 75% discount on input tokens. However, the key difference is that it isn’t automatic — users must use CachedContent.create to create a cache and explicitly specify usage when defining the model. offers a 75% discount offers a 75% discount Realtime API is last but not least. This feature could revolutionize speech-to-speech conversations, enabling seamless communication between humans and AI. Businesses could leverage it for customer service, scheduling, translation, or summarization. They could use it as an assistant, companion, or support tool. What These Updates Mean for ChatGPT’s Future The significance of focusing on API advancements and developer tools for this year’s DevDay cannot be ignored. Instead of rolling out a flashy new model at the big annual conference to make headlines and generate hype, OpenAI is ensuring its foundation is strong. These updates will encourage external innovation and generate insightful feedback from users. These updates could kick-start a growth period for OpenAI, causing speech-to-speech and image-focused models to trend. At the very least, other AI engineers and businesses will look toward this company as inspiration. Similar updates from competitor models will likely follow in the near future. As for the implications of Realtime API, prompt caching, vision fine-tuning, and model distillation, the one-day conferences in London and Singapore will likely be microcosms of what’s to come. Attendees and livestream viewers should expect to see developer-created content, insights, and questions featured at the upcoming events.

Stories, News & Mentions about OPENAI

Openai’s Stories on HackerNoon

Openai's News & Mentions