I may be an outlier, but I’ve worked at a startup company that did machine learning R&D, and which was recently acquired by a big tech company, and we did consider the issue seriously. The general feeling of the people at the startup was that, yes, somewhere down the line the superintelligence problem would eventually be a serious thing to worry about, but like, our models right now are nowhere near becoming able to recursively self-improve themselves independently of our direct supervision. Actual ML models basically need a ton of fine-tuning and engineering and are not really independent agents in any meaningful way yet.
So, no, we don’t think people who worry about superintelligence are uneducated cranks… a lot of ML people do take it seriously enough that we’ve had casual lunch room debates about it. Rather, the reality on the ground is that right now most ML models have enough trouble figuring out relatively simple tasks like Natural Language Understanding, Machine Reading Comprehension, or Dialogue State Tracking, and none of us can imagine how solving those practical problems with say, Actor-Critic Reinforcement Learning models that lack any sort of will of their own, will lead suddenly to the emergence of an active general superintelligence.
We do still think that eventually things will likely develop, because people have been burned underestimating what A.I. advances will occur in the next X years, and when faced with the actual possibility of developing an AGI or ASI, we’re likely to be much more careful in the future when things start to get closer to being realized. That’s my humble opinion anyway.
I’ve kept fairly up to date on progress in neural nets, less so in reinforcement learning, and I certainly agree at how limited things are now.
What if protecting against the threat of ASI requires huge worldwide political/social progress? That could take generations.
Not an example of that (which I haven’t tried to think of), but the scenario that concerns me the most, so far, is not that some researchers will inadvertently unleash a dangerous ASI while racing to be the first, but rather that a dangerous ASI will be unleashed during an arms race between (a) states or criminal organizations intentionally developing a dangerous ASI, and (b) researchers working on ASI-powered defences to protect us against (a).
What if protecting against the threat of ASI requires huge worldwide political/social progress?
A more interesting question is what if protecting against the threat of ASI requires huge worldwide political/social regress (e.g. of the book-burning kind).
I may be an outlier, but I’ve worked at a startup company that did machine learning R&D, and which was recently acquired by a big tech company, and we did consider the issue seriously. The general feeling of the people at the startup was that, yes, somewhere down the line the superintelligence problem would eventually be a serious thing to worry about, but like, our models right now are nowhere near becoming able to recursively self-improve themselves independently of our direct supervision. Actual ML models basically need a ton of fine-tuning and engineering and are not really independent agents in any meaningful way yet.
So, no, we don’t think people who worry about superintelligence are uneducated cranks… a lot of ML people do take it seriously enough that we’ve had casual lunch room debates about it. Rather, the reality on the ground is that right now most ML models have enough trouble figuring out relatively simple tasks like Natural Language Understanding, Machine Reading Comprehension, or Dialogue State Tracking, and none of us can imagine how solving those practical problems with say, Actor-Critic Reinforcement Learning models that lack any sort of will of their own, will lead suddenly to the emergence of an active general superintelligence.
We do still think that eventually things will likely develop, because people have been burned underestimating what A.I. advances will occur in the next X years, and when faced with the actual possibility of developing an AGI or ASI, we’re likely to be much more careful in the future when things start to get closer to being realized. That’s my humble opinion anyway.
I’ve kept fairly up to date on progress in neural nets, less so in reinforcement learning, and I certainly agree at how limited things are now.
What if protecting against the threat of ASI requires huge worldwide political/social progress? That could take generations.
Not an example of that (which I haven’t tried to think of), but the scenario that concerns me the most, so far, is not that some researchers will inadvertently unleash a dangerous ASI while racing to be the first, but rather that a dangerous ASI will be unleashed during an arms race between (a) states or criminal organizations intentionally developing a dangerous ASI, and (b) researchers working on ASI-powered defences to protect us against (a).
A more interesting question is what if protecting against the threat of ASI requires huge worldwide political/social regress (e.g. of the book-burning kind).