What is ChatGPT? and Other FAQs

What is ChatGPT?

Everyone should know what ChatGPT is and everyone should be using it. In fact, I believe that this wonderful OpenAI tool is revolutionizing industries. I use it at work every day, and will only use it more as time goes on. It’s not going anywhere.

Not everyone knows what it is, or has had cause to use it or learn how. I’ll address that first. Then, I’ll talk about ChatGPT and the rise of AI more in-depth. AI is changing the world forever. What we choose to do now will inform the direction of that change.

But I won’t get ahead of myself.

ChatGPT Chatbot

What is ChatGPT Chatbot and How Do I Use It?

ChatGPT is a Large-Language Model chatbot (LLM) based on OpenAi’s GPT-3 family of LLMs. Which is a lot of acronyms to say it is an Artificial Intelligence that can have conversations like a person.

ChatGPT was released in November of 2022. ChatGPT has taken the world by storm by writing accurately about a wide variety of topics. It is capable of writing well about everything from medicine to (older) geopolitics to art. It can even answer questions about convoluted board game rules or fantasy worlds. While it can be factually inaccurate, the program is getting better at communicating uncertainty as well as providing better answers. It’s also worth noting that it doesn’t have access to the internet yet, which is a critical limitation that will be improved in the future.

I’ve been using it since launch and I have been very impressed, to say the least. I will often ask questions that I would normally have to ask someone else or google and read web pages to answer. The responses are often excellent. It is usually easier to fact-check these responses than research something from scratch.

But, how do you use ChatGPT, exactly?

“Write me a poem about flowers.” or “Could you write me a poem about flowers?” or “Can you tell me about poems about flowers?”.

It’ll start typing a reply, and as long as you don’t overload its capacity to answer within a certain number of tokens (think of it as sort of like a character limit), it will type until it’s finished.

These replies are usually at least slightly different from each other, even for the same inputs. Of course, as a chatbot, it can remember previous inputs in a given conversation and incorporate them when responding to future inputs.

It’s pretty amazing. It even warns the user if it thinks it isn’t qualified to answer the question or will refuse to answer questions it deems inappropriate based on its training parameters.

ChatGPT: The Good

ChatGPT is obviously a boon to anyone who wants to write more, faster. I often find it easier to use ChatGPT to write 90% of something and then finish it yourself. Especially once you figure out how to write the best inputs for what you want. Use it to accelerate your ability to produce copy, reports, emails, or anything else that is even remotely rote.

It also works better than Google for certain functions, which is why it’s being incorporated into Bing. Even though it does produce errors, the app can’t access the internet directly, so it can’t fact check itself as effectively as it will in Bing. Still, it is an invaluable tool for answering questions that have a single consensus answer, in a fraction of the time it takes to use Google or another reference source.

As a chatbot, it can also be used to workshop and brainstorm. You can describe your problem or project, and then talk to it like you would any other colleague. Even if it doesn’t give you the answers you need, it can help you figure them out for yourself or give you ideas on where to look. It’s an unbelievably powerful tool.

A Chatbot That Can Handle Written Language Tasks

As a chatbot, it’s limited to using written language. Which means it can’t draw or paint. But, it can potentially write the code for a drawing or a painting. You can even use ChatGPT to create a prompt for a visual art AI like Midjourney. It can also describe a scene. There are people who are using it to write choose-your-own adventure books, after all.

ChatGPT Understands Written Language

Since it can understand written language, it can even code. It knows a variety of languages including, but not limited to: HTML, CSS, Javascript, PHP, Python, and more. I use it all the time to come up with easy solutions to minor development problems. I can’t stress this enough, it’s absolutely incredible.

This is also the source of some of its biggest problems. That brings us to my next point: the downsides and pitfalls of ChatGPT.

ChatGPT: The Bad

ChatGPT, AI, and the Future of Work

The most obvious problems with it stem from us coming to rely on it. People will lose their jobs. Technical skills that people have trained for years to master will become obsolete with each new iteration of this kind of software.

As a society, we will have to adapt to this change in the landscape. Of course, for now, people will have to input their needs and hold the hand of the AI. Expertise will lend itself to better hand-holding through better prompts and better oversight. But, in the future, this might not be the case. Which, depending on how we handle the shift in power, could be either good or bad.

Beyond that, it is becoming adopted very quickly. Once we come to rely on it, server outages and misinformation will have much farther-reaching impacts than just minor delays. A server error, downtime, or area of bad knowledge could result in a lot of lost time and money. Our reliance on AI could cause us to recover more slowly if we suddenly need to do something the old fashioned way.

I could also see a stagnation in innovation and original thought as the world becomes increasingly reliant on machines that are becoming trained on data they are at least partly producing.

Of course, these are mostly future concerns. This kind of technology is in its infancy and singularity isn’t here just yet.

The Limitations of ChatGPT

This means that the biggest problems presented by ChatGPT are its limitations. GPT can write serviceably. Formal or presentational writing needs to be reviewed and heavily edited to be up to par. Most of its creative endeavors are impressive for what they are, but are ultimately lackluster.

ChatGPT outputs often sound artificial and lack the x-factor imparted by a human with life experience, even in the most basic aspects of copy. The idiosyncrasies and flaws that make us human are not well-replicated. Sometimes this leads to lost time as I have to scramble to pull together a final project after ChatGPT has yielded mediocre results.

ChatGPT: The Ugly

On paper, and perhaps in retrospect, this is a watershed moment in human history. We stand on the precipice of a bold, terrifying, and powerful future of which we have collectively long-dreamt.

The thing is, when I thought about AI, I never thought about having to heavily edit or fact check. I didn’t think I’d be able to catch it in strikingly human inconsistencies or to be able to confuse it. It’s still in the early stages, but it makes me wonder what the potential for this technology really is. The realities of the market, labor, and politics are always stranger than one can predict. It’s a wonderful tool, but it is far from perfect. But, who knows what GPT4 or later versions will yield. Or even a dedicated writing bot. ChatGPT is ultimately trained to function as a chatbot, not a writer.

What to Use ChatGPT For

For now, my primary use for it is in writing outlines and the nuts-and-bolts of a piece which I then overhaul. It can make certain rote aspects of copywriting or blogging a lot faster and lets me accrue decision fatigue in other aspects of my work. Plus, this way, I am still adding significant value to products in a way that can’t be manufactured by a machine.

I also use it to fairly reliably synthesize, summarize, brainstorm, and draft more so than write. In other words, it is an excellent writing assistant and assistant writer, it has a long way to go before it achieves mastery.

I have no doubt it will one day replicate and surpass a lot of the work we do. But, in the meantime, there’s limitations and a wonky learning curve. I hope my thoughts have helped, and the FAQ that follows can get you started with your very own personal AI.

Chat GPT FAQ

Below is our FAQ on the OpenAI ChatGPT Chatbot and some basic tips on using this powerful tool effectively.

General ChatGPT Questions

Q: What is ChatGPT AI or ChatGPT Chatbot?

A: ChatGPT is an AI-powered chatbot created by OpenAI that can generate human-like text based on the input provided to it.

Q: What does "GPT" stand for in ChatGPT?

A: "Generative Pre-Trained Transformer”.

Q: Is there a ChatGPT app?

A: All you need to access ChatGPT is a web browser. While there are a number of apps that use ChatGPT to power them, none of those apps are made by OpenAI.

Q: Does ChatGPT save data?

A: Yes, ChatGPT saves data to improve the quality of its responses. However, OpenAI has stated that it takes user privacy very seriously and uses strict security measures to protect user data.

Q: How can I save a ChatGPT conversation?

A: You can save a ChatGPT conversation by taking screenshots or copying the text generated by the chatbot. Additionally, ChatGPT has a limited capacity to save previous conversations and the inputs/outputs within them. This way you can continue a conversational thread later.

Q. How much is ChatGPT Premium?

A. ChatGPT premium costs $20 a month. This subscription is well worth it as it provides the benefit of faster processing time (with DaVinci) and priority server access. Most of my prompts are replied to in seconds.

ChatGPT Technical Questions

Q: What is ChatGPT AI or ChatGPT Chatbot?

A: ChatGPT is an AI-powered chatbot created by OpenAI that can generate human-like text based on the input provided to it.

Q: What does "GPT" stand for in ChatGPT?

A: "Generative Pre-Trained Transformer”.

Q: Is there a ChatGPT app?

A: All you need to access ChatGPT is a web browser. While there are a number of apps that use ChatGPT to power them, none of those apps are made by OpenAI.

Q: I am getting an "internal server error" when using ChatGPT. What should I do?

Also sometimes asked as:
Q: What should I do if I get an "error in body stream" when using ChatGPT or a "network error" when using ChatGPT for long responses. What should I do?

A: These errors can occur due to network issues, server overload, or bugs in the software. If you encounter this error, try reducing the length of your input or splitting it into smaller parts. If the problem persists, try reloading the page or restarting the app. If that doesn’t work, contact the support team for assistance.

Q: Is there a character limit for using ChatGPT?

A: Yes, there is a character limit for using ChatGPT. The exact limit may vary depending on the app or platform you are using. ChatGPT is a large-language model (LLM) that functions off of tokens. Tokens roughly correspond to parts of words. More tokens are needed based on the complexity of the word or idea the words represent. There is a limit to how many tokens it will process at once.

Try to keep answers concise to avoid overloading the software. In the event that you cannot, try breaking prompts up into smaller questions if possible. This can often increase the quality of outputs.

ChatGPT for Education

Q: Can ChatGPT write essays?

A: Yes, ChatGPT can write essays on a wide range of topics. However, the quality of the essay depends on the input given to it and the parameters used for generating the text. Try creating an outline and making an input out of each piece. ChatGPT can even help you with this.

Q: Can ChatGPT make an essay sound better?

A: Yes, ChatGPT can generate text that can help make an essay sound better. However, it is important to ensure that the text generated is used ethically and in compliance with relevant laws and regulations. It’s best to use it to proof-read and suggest edits on existing text rather than to rewrite the text away from your own voice.

Q: Can I use ChatGPT to create a dataset?

A: Yes, you can use ChatGPT to create a dataset by generating text on a specific topic and saving the output. However, it is important to ensure that the dataset is used ethically and in compliance with relevant laws and regulations.

Want to discuss a Project?