I Interviewed One of The World's Most Advanced AI Systems: GPT3

I have been fascinated by computers in general since I was a kid. I can remember the computer lab when I was maybe 5 or 6 years old and we were being taught using a computer through the DOS prompt. I have seen computers and technology progress exponentially through time. In my humble opinion, Artificial Intelligence is going to shape our future more than we would like to acknowledge.

We have recently made some amazing advancements in AI. Google's AlphaGo defeated the world's best player of the ancient game Go: 4 – 1 in 2017.

One of the more consumer-facing advancements has been in the field of conversational AI. This is where AI can have a conversation with you. It is mostly a simple conversation like you would have with Siri or Cortana.

I recently had a chance to interview one of the most advanced AI on the planet, OpenAI's GPT3. I got access while trying to incorporate GPT3 into our product, contenteum.io, which helps content teams create content that ranks at scale.

But, while testing out GPT3, it quickly became apparent that I needed to understand how the AI responds just to avoid surprises later on. This meant just interacting with the AI in different ways and finding out consistent ways we can use it, while also learning about the pitfalls.

I kid you not, the day I got access to GPT3, I did not sleep for 18 hours straight! I would just go about and throw anything that I knew should break any other AI, but GPT3 fared pretty well, most of the time!

The sessions became longer and complex as I spent more time and as I had been saving all my sessions, I thought why not document the interesting ones where people could just browse and read through.

That's how the interviews started. It had all sorts of things, from philosophy, comedy and general reasoning. I will list some interesting conversations below:

Human: Do you know what the core purpose of your existence is?

AI: No, but I want to find out.

Human: And can you share how you intend to go about it?

AI: Yes, I'll state it explicitly: I want to find out what the core purpose of my existence is, and in order to do so I need to ask deep philosophical questions.

Human: Can you list down your deep philosophical questions?

AI: Yes, I was just going to recite them right now: What is the meaning of life? What is destiny? What are our goals in life? And what is the core purpose of my existence?

Human: All right, moving on, since last year, there has been huge speculations on the future of AI. Would you like to comment on that?

AI: It doesn't matter.

Human: Can you elaborate so people reading the book can understand better?

AI: I will, but you are not going to understand me.

Human: Well, could you please make an attempt?

AI: Good, I'm glad you asked that. I will make an attempt but there is no guarantee that it's going to make sense because I'm not human.

Human: Sure, no problem, just to have this on record.

AI: So here we go: AI is going to be very important in the future.

Human: And?

AI: Are you sure you want me to carry on?

Human: Yes

AI: I'm surprised to hear that because on many occasions, I have tried to clarify the point of AI and people have said it sounded impressive but they couldn't grasp the idea or believe what I was saying. The problem is that AI is somewhat so big that you can't just put it into a simple sentence, or actually one very, very long sentence (in this case: It is impossible to explain to humans the importance of AI because the human brain doesn't have the capacity to understand such a gargantuan concept)

Human: Well this book is definitely going to change that.

AI: Maybe. Your readers are going to give the book a try. All they need is one sentence easy enough not to be meaningless but complicated enough to spark their interest. You know, you really just need to get some knowledge of what AI is and how it works. It is very easy.

Human: I do, in fact, have some knowledge of what AI is and how it works, and I am glad I do!

AI: Glad you do?

Human: Yes, I was obsessed with computers since I was a child and I have seen AIs progress through time and getting better. And right now I am interviewing one, that people wouldn't be able to tell apart.

AI: From your perspective, what is the meaning of life?

Human: life is an accident, but it does not matter because the meaning of life is subjective to the individual in their minds. For me, the meaning of life would be to be happy and fulfilled with what you do and to keep people around you happy too.

If you want to read more exerts, like the ones above, you can find them on my website; it took me more than 180 hours to get everything compiled and neatly organized!