How close are the machines to taking over?

We overestimate the impact of technology in the short-term and underestimate the effect in the long run.

Amara’s Law (see https://fs.blog/gates-law/)

At one end of the debate we have people like Geoffrey Hinton flagging concerns about AI becoming able to control us. At the other end you’ve got people like Yann LeCun who tend to have a more optimistic outlook. Both people of similar levels of credibility in the space.

I’m going to suggest where I see the disconnect.

It’s in the language we use.

To most people, AI means something out of science fiction. Literally Skynet or I, Robot or Ex Machina: Something with its own motivations that are often at odds with humanity.

For researchers, the AI space is much broader. The NPCs you play against in computer games are AIs. You can even read about the AI behind the ghosts in the classic Pac-Man game. When AI researchers think about science fiction AI they use a different term: “Artificial General Intelligence” (AGI).

If you read that a researcher is talking about “AI” then you should be thinking: “wow, look how far we have come since Pac Man”. If they are talking about “AGI” then that is the beginnings of the path to science-fiction AI. But still just the beginnings.

I’ve made a handy graphic that shows where I think we are on this journey between Pac Man and Ex Machina. Obviously it’s somewhat tongue in cheek, but it’s informed by Amara’s law: There is a lot of hype about any new technology so people inevitably overestimate how much it will change things over the next year or two. But over the longer term …. a different story.