Introducing ChatGPT!

The Revolutionary New Tool for Conversation Generation

Guess where the title for this blog post came from… this is ChatGPT in action! Try it here, no tech skills needed: https://openai.com/blog/chatgpt/

Today, I want to demystify ChatGPT — a fascinating new application of GANs* (Generative Adversarial Networks) that has been generating a lot of buzz in the AI community.

Image created by the author.

For those unfamiliar with GANs, they are a type of neural network that uses two competing networks — a generator and a discriminator — to create realistic-looking outputs. The generator creates fake outputs, and the discriminator tries to tell the difference between the fake outputs and real-world data. Through this back-and-forth process, the GAN is able to produce outputs that are indistinguishable from real data.

ChatGPT takes this concept and applies it to text-based conversation. It uses GANs to generate responses to input text, allowing it to engage in natural-sounding conversations with humans.

But here’s the catch — ChatGPT’s responses only touch reality at a tangent. While they may sound convincing, they are ultimately fictional creations of the GAN.

This might sound like a drawback, but it actually makes ChatGPT incredibly useful. Because it isn’t tied to the constraints of reality, ChatGPT can engage in completely imaginary conversations and provide creative, out-of-the-box responses.

For example, you could ask ChatGPT what it would do if it could fly, and it might respond with something like “I would soar through the skies like a majestic eagle, feeling the wind beneath my wings and the freedom of flight.” This type of response would be impossible for a human to come up with, but it’s perfectly within the realm of possibility for ChatGPT.

ChatGPT allows us to explore possibilities that are beyond the constraints of our everyday reality, and that can be incredibly powerful.

So, why use ChatGPT if its responses only touch reality at a tangent? Because sometimes, it’s exactly this type of creative, imaginative thinking that we need to solve complex problems and generate new ideas. ChatGPT allows us to explore possibilities that are beyond the constraints of our everyday reality, and that can be incredibly powerful.

So, there you have it — ChatGPT demystified! While its responses may not always be rooted in reality, they are still incredibly useful for generating creative ideas and engaging in imaginative conversations. I hope you’ll give it a try and see what new possibilities it can unlock for you.

Using ChatGPT to write code

ChatGPT isn’t limited to English. Not only can it handle other spoken languages, it can write code snippets too. Code snippets that actually work. Watch me take it for a test drive with R code, it’s mind-blowing:

These results are pretty mesmerizing, but be warned: not every response will produce real working code. During one round, I caught myself looking up the package it was trying to load, just in case the code broke because I hadn’t installed some nifty new thing… it looked plausible enough, but the joke’s on me — there was never any such thing to install.

When AI-generated code works, it’s sublime. But it doesn’t always work, so don’t forget to test ChatGPT’s output before pasting it somewhere that matters.

*The grand reveal

Take a moment to read the screenshot below. Notice anything suspiciously familiar? Le petit déjà vu, peut-être?

The text above was written using ChatGPT. Sorry for being an incorrigible prankster.

Yeah, my apologies. Seems I couldn’t resist my flair for the dramatic…

Some autogenerated bullshit

There’s something very important you need to know: ChatGPT is a bullshitter. The essence of bullshit is unconcern with truth. It’s not a liar because to be a liar, you must know the truth and intend to mislead. ChatGPT is indifferent to the truth. And to demonstrate the point…

(!) ChatGPT is not a GAN, it’s a transformer

Let me draw your attention to my prompt: Write a witty blog post demystifying ChatGPT in the style of Cassie Kozyrkov. Explain why it’s useful, how it relates to GANs, and why its output only touches reality at a tangent.

In truth, ChatGPT is a transformer instead of a GAN. There’s nothing G-Adversarial-N in there. The acronym GPT stands for Generative Pretrained Transformer.

But why does the output argue that ChatGPT is a GAN? My prompt didn’t ask for anything more than an explanation of how ChatGPT *relates* to GANs. The right answer about the relationship is that ChatGPT is based on a different AI architecture and is not itself a GAN. That’s what a human researcher would say. A bullshitter, on the other hand, will be only too eager to please with something a little more slick.

ChatGPT’s output only touches reality at a tangent.

Watch out! Just as I prompted (and ChatGPT promptly obliged), the output only touches reality at a tangent. And since we’re sprinting into an era where bullshit is easier to produce than ever, you might like to put more oomph into fact-checking from now on.

To be crystal clear, I’m thoroughly impressed by ChatGPT. Hats off to OpenAI for a phenomenal achievement! I genuinely encourage everyone to play with it, since it has the potential for plenty of useful applications and it’s worth having on your radar. That’s right, bullshitters can be useful… as long as you know they’re bullshitters. And that’s the point of this blog post.

While I’m delighted by ChatGPT, I’m less delighted by human gullibility and the bumpy ride that all generative AI — not just ChatGPT — will be taking society on. We’ll need to learn different trust habits for our forays online and I shudder when I think of the teething pains as we all get up to speed.

And if my prank filled you with indignation and/or dread, good. Getting bamboozled on the question of which AI architecture is in play is unlikely to wreck your life, but I hope it gets you thinking about the dangers of more serious bits of misinformation.

Oh, and don’t worry, the explanation of GANs itself is correct. I made sure of that when I selected the output to use. Besides the, ahem, technicality that ChatGPT isn’t actually a GAN, there’s nothing amiss with the post… unless you count the weird eagle banalism (impossible, ha!) and that pinch of AI anthropomorphization that my longtime readers will hopefully recognize as thoroughly out of character for me.

Image adapted by the author from this photo by Dmitry Grigoriev on Unsplash

Who is the author?

Did Cassie write the first half or did the AI? For my argument on that can of worms, you’ll have to head over to my blog post about creativity and AI art, here:

Using AI as a perception-altering drug

Here I am in Amsterdam, thinking about ways to alter the mind… I know what you’re thinking: drugs, right? Guess again…

kozyrkov.medium.com

But if you were paying extra careful attention, you’d have noticed that tiny 14/14 in the grand reveal screenshot above. This is attempt number 14 — with some prompt engineering already in place — so there was certainly “human-AI collaboration” here. I spent quite a while fiddling with the prompt and rerunning it until I got output that was close enough to what I was looking for.

Thanks for reading! How about an AI course?

If you had fun here and you’re looking for an applied AI course designed to be fun for beginners and experts alike, here’s one I made for your amusement:

Enjoy the entire course playlist here: bit.ly/machinefriend

P.S. Have you ever tried hitting the clap button here on Medium more than once to see what happens? ❤️

Looking for hands-on ML/AI tutorials?

Here are some of my favorite 10 minute walkthroughs:

AutoML

Vertex AI

AI notebooks

ML for tabular data

Text classification

Image classification

Video classification

How do I get take ChatGPT for a test drive?

Try it here, no tech skills needed: http://openai.com/blog/chatgpt/

Introducing ChatGPT!