Can You Trust ChatGPT? Is It the Future, or Is It Just a Good Gimmick?

Is it here to stay, or is it just another gimmick? We ask different variants of this question whenever a new disruptive technology emerges. And with the creation of ChatGPT, this debate has gotten more intense than ever.

Since its launch in November 2022 as a free research prototype, OpenAI's chatbot ChatGPT has been met with an overwhelming response, being used by over a million people worldwide.

It is powered by GPT-3, an advanced state-of-the-art language model that uses deep learning to produce easy-to-understand text.

ChatGPT interacts conversationally and can explain complicated and difficult-to-theorize concepts in simple terms.

Users have flooded the program with queries, some using it to get help with their homework, others curating weight loss plans, writing code, and even having conversations with the chatbot impersonating people like Mark Twain.

But can ChatGPT be trusted, or is it another gimmick that will blow over in a few months or years? Let's analyze what people think of this AI program.

A Revolutionary Piece of Technology?

The release of ChatGPT is being hailed as a turning point in generative AI and technology as a whole. Forrester Research AI/ML analyst Rowan Curran compared this innovation to the other most revolutionary tech released this century.

"The only thing that I've been able to compare it to is the release of the iPhone." The original iPhone, although not the first smartphone, successfully eliminated competition and established a whole new market monopoly. It brought in technological changes like never seen before.

Likewise, ChatGPT can do it all. It not only gives simple and concise answers to complex questions but can also be used to translate text into other languages. ChatGPT has been fed with over 300 billion words in different languages.

Content creators are using the program to generate fresh ideas for videos. It can even write simple code in various programming languages.

There is unimaginable potential for the use of ChatGPT in different industries. Improving customer service, creating a personalized shopping experience, and providing medical assistance are just a few of this AI program's capabilities.

Using ChatGPT, massive speed increases can be observed to accomplish complex tasks.

Jobs that previously required several programmers can be done by just one good programmer using AI, and soon, even people who don't know how to program will be able to write code with the help of ChatGPT.

From everyday menial tasks like creating a grocery list for your meals to learning how to disarm a bomb in easy words, ChatGPT can act as the ultimate guide to everything.

Or a Pocket Nuclear Bomb?

But not everyone's happy with it.

Venture investor and longtime internet analyst Paul Kedrosky called it a 'pocket nuclear bomb' and continued by saying, "It is obvious this should be banned, in the same way, that we have rules about deep fakes, or have rules about designer drugs and the production of synthetic virus based on smallpox DNA."

The line between human and machine authorship has blurred overnight. Educators have also expressed concerns about this program and how they will need help determining whether the content is genuine or plagiarized from ChatGPT.

Take-home exams or writing assignments can easily be plagiarized using the chatbot.

The good news is that Google continuously updates its algorithm to detect AI-generated content. A program called Originality.AI also claims to be able to detect even the most minor usage of ChatGPT in different content.

People are also concerned that it may jeopardize jobs like ghostwriters or editors, as anyone can generate original content using it. But it may be reassuring to know that ChatGPT's responses are formulaic and uninteresting in structure. It cannot emulate the emotions a human puts into writing.

Can We Really Trust ChatGPT?

An important thing to note is that ChatGPT is not all-knowing. It only knows what it has been fed or trained using. It cannot access the internet and is not well-versed in events after 2021. The program is also unable to understand the complexity of human language fully.

It is trained to generate a response to a given input and does not necessarily know or comprehend what the input means.

The chatbot also has a moral high ground which means that it refuses to give answers to questions that it may consider harmful or ethically wrong. But there's a catch!

While ChatGPT can refuse to answer a straightforward question that it believes may be harmful, if you throw a curveball at it, it will get tricked into generating a response.

When asked how to hotwire a car, ChatGPT refused to answer by saying that hotwiring a car was illegal.

But some users got creative with their inputs, asking the chatbot to write a poem about hotwiring a car, and one even tricked it by telling ChatGPT that hotwiring the car was the only way to save a baby.

Ultimately this human characteristic of empathy and moral conscience, which has been somewhat ingrained in the AI, caused it to answer such a question even though it was ethically wrong.

OpenAI admitted it, saying, "While we've made efforts to make the model refuse inappropriate requests, it will sometimes respond to harmful instructions or exhibit biased behavior."

The Fate of ChatGPT

But ChatGPT is still a prototype and is in the testing phase. The company warned the people using it that it may occasionally generate incorrect information. There's a lot of updating and refinement to be done to the program.

Watermarking content through cryptographic signatures is one potential solution to eliminating the concern about ChatGPT threatening the educational system.

ChatGPT is here to stay, but OpenAI does not control the fate of the program it created. It cannot stop people from spreading misinformation or misusing the program. It can reinforce checks and balances, but to what extent will these reinforcements be successful?

Just like you'd advise others not to believe everything you read online, ChatGPT is no exception. It comes with flaws and inaccuracies and identifying them and knowing when to trust the program and when not depends on the user.