Timestamp: GenML + ChatGPT, etc.

May 24, 2023

So it's been five months since I put a time stamp against all the Metaverse hype and now it's time for everything Generative ML.

Sam Altman is on a PR tour. Every podcast is interviewing him or talking about him and ChatGPT 4. I'm getting a couple of messages a week from friends outside of tech asking if I've heard of it and just generally loosing their shit about it.

I have a tab open in Chrome to read later where Bill Gates is claiming AI is as big a deal as mobile phone and the internet.

Most of the conversations I'm having tend to go like this:

1. ChatGPT is very impressive, but it is trained on 100% exisiting information and therefore can't come up with anything new
2. How is that different to how our brains work?
3. What is an original idea?
4. What is consciousness and does that even matter here?

The bear case is something like, yeah it's going to be able to pass all the entrance exams and summarise the storyline of Dragon Ball Z and write production ready code in seconds, but we will always have to tell it what to do. It will never know what to do. Or even want to do anything.

The bull case is that this is a significant stepping stone towards AGI and we are witnessing the beginning of the next fundamental shift in technology that will change...everything.

Altman made an interesting point this week when thinking about a future where we can engulfed in AI products. It's been over 20 years since we declared that no human would ever beat a computer at chess again and yet, we still watch two humans competing and not two AI programs. And chess is probably more popular than at any time in history.

I agree that making any kind of predictions here is crazy considering the rate of change.

It certainly feels that lots of our day-to-day tasks and roles will be replaced or changed in the next couple of years. But, (and I'm aware I'm biased to think this) it does feel like there is something baked into being human that feels bit chaotic and silly and fun. We make mistakes and have playful; ideas that turn into art, poetry, music, products, and more.

We feel things and need to do something with those feelings or we're likely to explode. I don't know if computers feel things, I doubt it. I'll check...

Okay, phew. They don't.

It's impressive that a program can read a lot of books and articles and then predict with a very good level or accuracy what word I'm going to type next to the point that it feels like it's mimicking a human. But it still doesn't know what a word or a human is. And maybe it never will?

Either way, technology feels exciting again and it's fun to watch, for now.