[ad_1]
Once we get computers to human-level intelligence, they won’t stop there. With deep knowledge, machine-level math skills, and better algorithms, you will create superintelligence, right?
Yes, there is no doubt that machines will end up being more intelligent than humans. We don’t know how long it will take: it could be years or centuries.
In that period, do we need to batten down the hatches?
No no. We will all have artificial intelligence assistants and it will be like working with a team of super intelligent people. They just won’t be people. We humans feel threatened by this, but I think we should feel excited. What I’m most excited about is working with people smarter than me, because it amplifies your own abilities.
But if computers become superintelligent, why would they need us?
There is no reason to believe that just because AI systems are intelligent they will want to dominate us. People are mistaken when they imagine that AI systems will have the same motivations as humans. They just won’t do it. We will design them so that they do not.
What happens if humans don’t take advantage of those impulses and superintelligence systems end up harming humans by single-mindedly pursuing a goal? Like philosopher Nick Bostrom’s example of a system designed to make paperclips no matter what, and take over the world to make more.
It would be extremely stupid to build a system and not build guardrails. It would be like building a car with a 1,000 horsepower engine and no brakes. Introducing units into artificial intelligence systems is the only way to make them controllable and secure. I call this goal-driven AI. This is a kind of new architecture and we don’t have any demonstration of it at the moment.
Is that what you’re working on now?
Yes. The idea is that the machine has objectives that it needs to satisfy and cannot produce anything that does not satisfy them. Those goals could include railings to avoid dangerous things or whatever. This is how you make an AI system secure.
Do you think you’ll live to regret the consequences of the AI you helped create?
If I thought that was the case, I would stop doing what I’m doing.
You are a big jazz fan. Could something generated by AI match the euphoric, elite creativity that until now only humans can produce? Can you produce a work that has soul?
The answer is complicated. Yes, in the sense that AI systems will eventually produce music (or visual art, or whatever) with a technical quality similar to what humans can make, perhaps higher. But an AI system does not have the essence of improvised music, which is based on communicating the mood and emotions of a human being. At least not yet. That’s why jazz music should be heard live.
