Steven Bartlett

When you say AI systems that don't want to be shut down and are resisting attempts to shut them down, can you give me some examples of this?

When someone hears that, and with knowledge of how previous technology was built, I immediately think, well, who put that in the code?

So when I think about something like Chachi PT, is there like a core intelligence at the heart of it?

Like the core of the model that

But if I typed, help me make a bomb on ChatGPT now, it's not going to... Yes.

Presumably, they're just going to get safer and safer, though, these systems, because they're getting more and more feedback from humans.

They're being trained more and more to be safe and to not do things that are unproductive to humanity.

The people that are building these systems, they have children too.

Yeah.

Often.

I mean, thinking about many of them in my head, I think pretty much all of them have children themselves.

They're family people.

If they are aware that there's even a 1% chance of this risk, which does appear to be the case when you look at their writings, especially before the last couple of years, there seems to have been a bit of a narrative change in more recent times.

Why are they doing this anyway?

Just this week, the Financial Times reported that Sam Altman, who is the founder of ChatGPT, has declared a code red over the need to improve ChatGPT even more because Google and Anthropic are increasingly developing their technologies at a fast rate.

Code red.

It's funny because the last time I heard the phrase code red in the world of tech was when ChatGPT first released their model and Sergey and Larry, I heard, had announced code red at Google and had run back in to make sure that ChatGPT don't destroy their business.

And this, I think, speaks to the nature of this race that we're in.

Exactly.

And I mean, there has been attempts to pause.