Everyone knows it's terrible, but it's cool that they could do this at all. Like, it's a small preview of what's possible. Imagine this 10 years from now.

10 years? Imagine it 2 years from now. Or even tomorrow, because it's unlikely this is the apex of what's possible now.

I imagine in 10 years it'll be the same. Firstly because if you can't get "enough" good content to model, you will have to manually shape the models to make it sound authentic - and that isn't AI, that's editing. But secondly because humans are very good at recognizing authentically human interactions, and algorithms simply aren't good enough to generate all the nuances we identify as authentic.

If you want to fake someone out, use a regular person, and clean it up in post.

Yeah. The first few seconds of Rogan talking were pretty decent, but Jobs doesn't sound right at all, mainly because of the dialogue specifically.

But I do kind of get the hype.

The first few minutes sounded bad but the last half was quite impressive.

The first minute or two of Jobs is pretty bad, but after that it is not that bad. Not totally convincing, but it isn't bad. I suspect most people saying this did not listen past that part.

It's state of the art. It sounds amazing compared to anything that's existed before. What alternative are you comparing this to to say that it's awful?

A contrarian outburst seems to be the default on HN these days. Perhaps it has seeped in via the rampant infestation of politics. The ‘hackers’ get blasted when sharing projects.

Personally I’m in awe of this tech. But I also detest its inevitable weaponisation. But the tech! Wow.

If you are only looking at the current product and not the potential that comes with it then you are missing the point.

The problem is - can the potential ever be realised?

Self-driving cars are, what, 90% of the way there? And have basically been stuck at "almost there" for quite some time.

AR had huge potential when I first played with it in the early 2000s. Still basically nowhere.

So we've now got systems which can generate text which fools people into thinking it was written by a human - well, Eliza got there first.

Text to speech? The Voder was doing that in 1939! This stuff is better, sure, but pretty far from being realistic.

Perhaps this will succeed. But potential doesn't always mean achievable.