Godfather of AI Says There's an Expert Consensus AI Will Soon Exceed Human Intelligence

Geoffrey Hinton fears that we aren't doing enough to ensure the safe development of AI, with military applications posing the biggest threat.

4 comments

Eventually...

The headline uses "soon" while the article (being generous there) quotes him with h "eventually"
- Clickbait!
  
  Also the article mixes several distinct concepts that cannot be lumped together. The impending robot apocalypse like in the old scifi/action movies, ChatGPT, AGI and narrow AI that enables murder drones aren't the same thing. And if they are, I'd like to also add "algorithms" and recommendation systems making us more stupid and disuniting society.
  
  I'm not saying it doen't need to be addressed, but the argumentation is just flawed.
  
  There is one paragraph I completely agree with: »He's "worried about AI taking over mundane jobs." This would boost productivity, Hinton added, but the gains in wealth would disproportionately go to the wealthy [...]«
  
  I'd say there is a 100% chance of this happening, unless someone steps in and regulates things. That's the one thing I find nice of Mark Zuckerberg (and a few others), that he gives state of the art language models to the people and enables them to take part in AI at all. Other than that, it's a game that is entirely directed and controlled by the rich and large tech companies. And we shouldn't rely on them (or Mark Zuckerberg) shaping our future.
Well, half of the us still supports trump, so it's not like it's a high bar or anything.
That's a pretty low benchmark.