\
In humanity's latest installation of groundbreaking machine intelligence, ChatGPT—a conversational chatbot that provides a unique answer to any question you ask—seems to have outdone itself. According to the creators of the viral chatbot, [ChatGPT](https://hackernoon.com/chatgpt-is-a-plague-upon-online-publications), boasts the ability to __["answer followup questions, admit its mistakes, challenge incorrect premises, and reject inappropriate requests"](https://openai.com/blog/chatgpt)__. That is extremely impressive, considering how easy it had been to stump its less illustrious predecessor just a couple of years ago. Now, a mere generation later, many have tried and failed in this endeavor, promptly sparking excited chatter in closed circles over ChatGPT's potential application across diverse fields and industries.

\
Undeniably, ChatGPT is good. But is it really good for everything?

## What Is ChatGPT good for?

Built as an all-purpose chatbot, ChatGPT is really good at what it's designed to do: answering questions. As demonstrated by hordes of netizens on social media, ChatGPT's vast knowledge about the world allows it to generate responses ranging from __[complicated bug fixes](https://www.pcmag.com/news/watch-out-software-engineers-chatgpt-is-now-finding-fixing-bugs-in-code)__ to __[writing rap lyrics for a Drake song about beans](https://www.tiktok.com/@gengarcade/video/7174893425576168746?lang=en)__. In fact, unless the request infringes upon ethical boundaries set by its creators, it seems as if no question is ever too weird or niche for ChatGPT to take on.

\
On top of answering mundane questions predominantly undertaken by Google—sometimes with a surprising dash of humor—one thing that sets ChatGPT apart from search engines is its ability to distill information and provide answers to very specific questions. 

\
**Consider the following user prompt**: *I am an AI engineer with 5 years of software development experience seeking a role in machine learning research at MIT. Can you give me an example of a good cover letter that will impress my interviewers?* 

\
Insert that chunk into [Google](https://hackernoon.com/company/google) and it returns generic results about how to write cover letters and some examples of similar search queries.

\
*  ![Asking ChatGPT to write a cover letter for me \\ Source: author's own screenshot](https://cdn.hackernoon.com/images/Qp9ftgG0hXTRbuACmzH3L9QF5Qi1-2023-03-16T13:45:57.171Z-clfb5zd83001t0bs69s65ay0o)

\
In comparison, ChatGPT writes you a customized cover letter that includes the information you have provided in your input, saving you time from scouring through dozens of cover letter examples on search engines, with no guarantee of finding a suitable one. 

\
For millions of busy folks worldwide, ChatGPT's ability to drastically shorten research time makes it a far superior choice to current search engines. Consumer demand speaks for itself: Globally, the Bing app (now ChatGPT-powered) __[saw a 10x jump in new downloads](https://techcrunch.com/2023/02/08/bings-previously-unpopular-app-sees-a-10x-jump-in-downloads-after-microsofts-ai-news/)__ less than 24 hours after Microsoft's surprise announcement. Months into 2023, the war is already on.

\
## **ChatGpt: Good but not good enough**

I'm not being insincere when I say that ChatGPT is good. Objectively speaking, ChatGPT fulfills all the criteria of an excellent chatbot: it gives customized and complete answers, has a fast response time and is extremely knowledgeable about a wide variety of topics. No matter what universe you live in, ChatGPT sounds like the perfect chatbot. But that's precisely the problem—it's simply a very good chatbot, and not everything cannot be done on a chatbot. Some of ChatGPT's strengths backfire when you need to get down to serious business:

### Dialogue ≠ Editor

One of the factors that contribute to ChatGPT's immense popularity is its ease of use. Engaging with ChatGPT is so easy that users simply have to talk to the chatbot in a way one would text a friend to ask a question. You're not required to type in complete sentences, nor do you have to constantly remind the chatbot what you guys have been chatting about. This makes asynchronous communication extremely convenient, and users are able to resume their conversations from where they left off whenever they want to.

\
On the flip side, this makes it difficult for serious writers to use ChatGPT. Whether you're writing a listicle, how-to guide, or comparison article, there’s a fixed structure to adhere to. Dialog format, while catchy and requires minimal effort, is in no way built to support blog writing—the lack of an in-built editor makes it impossible to write long-form content with ease and instead requires you to switch back and forth between chat and your working document. While this may seem doable for the first few articles, it quickly becomes annoying when what should have been a straightforward workflow spirals into a chat-edit-search multi-tasking madhouse.

\
So, unless you have three separate monitors that you can refer to at any one time, ChatGPT may not offer the best workflow for serious content creators.

\
### **Unplugged & Uninformed**

Much like human-to-human interactions, the content that ChatGPT provides comes with a huge caveat: it isn't connected to the internet. This means that the chatbot provides responses using only internally-stored data that makes up its language model, which is, while impressive on its own, not a good thing for users because the chatbot cannot fact-check external sources or retrieve more in-depth information for you. 

\
This would have been fine if you were a WWII historian checking up on facts and figures, but for work that relies on ongoing trends and up-to-date sources, importing the content provided by ChatGPT wholesale just won't fly. Worst case scenario, you might get called out by the entire Internet for spreading fake news. __[Just ask Bard how that feels.](https://www.theverge.com/2023/2/8/23590864/google-ai-chatbot-bard-mistake-error-exoplanet-demo)__

\
 ![Source: imgflip.com](https://cdn.hackernoon.com/images/Qp9ftgG0hXTRbuACmzH3L9QF5Qi1-2023-03-16T13:45:57.172Z-clfb5zd84001u0bs6apan39uh)

\
Think of the chatbot as that one smart friend you have who seems to know everything, only slightly worse. At the very least, your friend reads the news every day and can hold conversations about what's happening around the world right now. 

\
In comparison, because ChatGPT was trained using only text data up to September 2021, it is oblivious to anything that happened after. Perhaps a better analogy then would be thinking of ChatGPT as that smart friend who decided to move into a cave during a zombie apocalypse—an apt analogy, because ChatGPT knows nothing about our (post-)pandemic world.

## Unabashedly, Utterly Wrong

By now, it's no secret that these seemingly all-knowing chatbots make amateur blunders as well. According to its creators, ChatGPT has a tendency to __["sometimes write plausible-sounding but incorrect or nonsensical answers"](https://openai.com/blog/chatgpt/)__, a well-documented phenomenon known colloquially as "hallucination" among people within the AI circle. Even when we know it’s wrong, ChatGPT responds so convincingly and confidently that sometimes makes us wonder if we were actually mistaken.

\
 ![ChatGPT apologizing for citing non-existent research papers, promises to write a summary based on existing papers, then proceeds to cite two more non-existent papers | Source: economistwritingeveryday.com](https://cdn.hackernoon.com/images/Qp9ftgG0hXTRbuACmzH3L9QF5Qi1-2023-03-16T13:45:57.173Z-clfb5zd85001v0bs62hks4in1)

\
\
Granted, many of these mistakes do not involve high-stakes scenarios where a single error in an ad campaign has led to life-threatening consequences. But if anyone has learned their lesson from the recent __[$100 billion Bard slip-up](https://www.npr.org/2023/02/09/1155650909/google-chatbot--error-bard-shares)__, it's that being accurate and factual means a lot in content marketing, especially if you want your customers to believe you know exactly what you're doing. Until the issue gets fixed—which, according to experts, remains challenging to solve—the onus is still on content creators to ensure complete accuracy in the content they put out, and relying on ChatGPT to get the job done is a risky proposition that might backfire on your business if you're not too careful.

\
I can guarantee you won’t want to be the one caught spreading misinformation through ChatGPT.

## Not ChatGPT, but something else?

Chatbots and the technology that powers them have existed for decades. Like many large language models out there, these models are merely trained to predict the likelihood of utterances, which says nothing about the accuracy of the information they generate—something that is extremely crucial in my line of work. So imagine my disappointment when I finally hopped onto the hype bandwagon only to realize that ChatGPT was more gimmicky than functional.

\
But that’s not to say that all AI tools are useless. In my __[determined search](https://www.forbes.com/sites/markminevich/2023/01/29/the-generative-ai-revolution-is-creating-the-next-phase-of-autonomous-enterprise/?sh=4a4be85c1bc1)__ for AI content marketing tools to help make my life easier, I chanced upon a platform that actually delivers what it sells. Unlike ChatGPT, their AI tools have built-in capabilities that allow them to scrape URL links and read content from webpages. To my pleasant surprise, I discovered that they also very recently built their own chatbot—one that actually returns sources from the Internet instead of “hallucinating” its own. They are called Hypotenuse AI —go check them out!

The writer is smart, but don't just like, take their word for it. #DoYourOwnResearch

Read My Stories

Too Long; Didn't Read

Build an AI-based text classifier with Tensorflow & NoSQL

How I Feel About ChatGPT As A Content Marketer

Too Long; Didn't Read

@lemonsqueeze

Credibility

RELATED STORIES