Article directory
Introducing ChatGPT!
Original: Address
Author: Cassie Kozyrkov
Chief Decision Scientist, Google. ❤️ Statistics, ML/AI, Data, Puns, Art, Drama, Decision Science.
Introducing: A Revolutionary New Tool for Conversation Generation
Guess where the title of this blog post came from... Here's ChatGPT in action!
Try it here, no technical skills required:
https://openai.com/blog/chatgpt/
Against Network GANs
Today, I want to demystify ChatGPT - a fascinating new application of GANs* (Generative Adversarial Networks) that is causing quite a stir in the AI community.
For those unfamiliar with GANs, they're 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, while the discriminator tries to tell the difference between fake outputs and real data. Through this back and forth process, GANs are able to produce outputs that are indistinguishable from real data.
ChatGPT takes this concept and applies it to text-based conversations. It uses GANs to generate responses to input text, enabling it to have natural-sounding conversations with humans.
But here's the catch - ChatGPT's response only touches reality on a tangent. While they sound convincing, they are ultimately fictional creations of GANs.
This may sound like a drawback, but it actually makes ChatGPT very useful. Because it is not bound by reality, ChatGPT can hold fully imagined conversations and provide creative, out-of-the-box responses.
For example, you can ask ChatGPT what it would do if it could fly, and it might respond like this: "I will soar in the sky like an eagle, feel the wind under my wings and the freedom to fly." It is impossible for humans to come up with this type of response, but it's well within the possibilities of ChatGPT.
ChatGPT enables us to explore possibilities beyond the constraints of our everyday reality, which can be very powerful.
So why use ChatGPT if its responses only touch reality tangentially? Because sometimes, that creative, imaginative mind is what we need to solve complex problems and generate new ideas. ChatGPT enables us to explore possibilities beyond the constraints of our everyday reality, which can be very powerful.
So there you have it - ChatGPT Demystified! While its responses may not always be grounded in reality, they can still be very useful for generating ideas and having imaginative conversations. I hope you give it a try and see what new possibilities it unlocks for you.
Write code using ChatGPT
ChatGPT is not limited to English. Not only can it handle other spoken languages, but it can also write code snippets. Code snippet that actually works. Watch me test drive it with R code, it's amazing.
These results are fascinating, but be warned: not every response will result in actual working code. In one round, I found myself looking up the package it was trying to load, just in case the code broke because I didn't install some fancy new stuff...it seemed reasonable, but I'm kidding - never did stuff stuff to install.
When AI-generated code runs, it's sublime. But it doesn't always work, so don't forget to test ChatGPT's output before pasting it somewhere important.
big reveal
Take a moment to read the screenshots below. Notice anything suspiciously familiar?
The text above was written using ChatGPT. Sorry, I'm a hopeless prankster.
yes i'm sorry. Seems like I can't resist my flair for drama...
some auto generated crap
There is one very important thing you need to know: ChatGPT is bullshit. The essence of bullshit is a lack of concern for the truth. It's not a liar because to be a liar you have to know the truth and intend to mislead. ChatGPT is indifferent to the truth. And prove it...
(!) ChatGPT is not a GAN, it's a transformer
Let me draw your attention to my tip: write a witty blog post demystifying ChatGPT in the style of Cassie Kozyrkov. Explain why it works, how it relates to GANs, and why its output only touches reality at tangents.
Actually, ChatGPT is a transformer rather than a GAN. There isn't any G-Adversarial-N out there. The acronym GPT stands for Generative Pre-Training Transformer.
But why does the output show that ChatGPT is a GAN? My tip only asks to explain how ChatGPT relates to GAN. The correct answer about the relationship is that ChatGPT is based on a different AI architecture, which itself is not a GAN. That's what a human researcher would say. A braggadocio, on the other hand, will be very eager to please with something smoother.
ChatGPT 的输出只是切线地触及现实。
careful! As I was prompted (and ChatGPT immediately obliged), the output just hits reality on a tangent. Since we're sprinting into an era where bullshit is more likely to be produced than ever, you might want to start investing a little more effort into fact-checking now.
To be clear, I am very impressed with ChatGPT. Hats off to OpenAI for a remarkable achievement! I sincerely encourage everyone to use it, as it has the potential for tons of useful applications and deserves your attention. Yes, braggarts are useful...as long as you know they are braggarts. That's the point of this blog post.
While I'm happy for ChatGPT, I'm less happy about human credulity and the bumpy journey all generated AI - not just ChatGPT - will take society. We need to learn different habits of trust for our online attempts, and I shudder when I think about the teething pain we're all picking up speed.
If my pranks make you angry and/or terrified, that's fine. Getting obsessed with the question of which AI architecture is at work is unlikely to spoil your life, but I hope it gets you thinking about the more serious dangers of misinformation.
Oh, don't worry, the explanation of GAN itself is correct. I determined this when I selected the output to use. There's nothing wrong with this post other than the technical issue that ChatGPT isn't actually a GAN...unless you count the weird eagle cliché (impossible, ha!) It fits my personality.
Image adapted by the author from this photo by Dmitry Grigoriev on Unsplash
Who is the author?
Was the first half written by Casey or by the AI? For my argument about that bunch of worms, you have to go to my blog post on creativity and the art of AI, here:
Using AI as a perception-altering drug
I'm in Amsterdam, thinking about ways to change minds... I know What were you thinking: drugs, right? guess again...
But if you pay close attention, you'll notice the small 14/14 in the larger screenshot above. This is the 14th attempt - with some timely engineering in place - so there's definitely "human-machine collaboration" here. I spent a long time playing around with the prompt and re-running it until I got output that was close enough to what I was looking for.