Ask A Genius 894: Hype on Self-Driving Cars
Author(s): Rick Rosner and Scott Douglas Jacobsen
Publication (Outlet/Website): Ask A Genius
Publication Date (yyyy/mm/dd): 2024/02/05
[Recording Start]
Rick Rosner: All right, this is an article from today’s LA Times, January 5th, 2024: Calling out the hype around AI, self-driving cars by Michael Hiltzik, and it’s an article about this guy named Rodney Brooks, who started Roomba and claims to have built more robots than anybody else and it’s Rodney Brooks’s sixth annual predictions scorecard. He looks at various tech areas and predicts how they will go in the future because it’s dumb to make predictions about the past. He agrees with Cory Doctorow that there’s way too much hype about AI and that, in the near future, it won’t live up to the hype. He says that he doesn’t think that AI with well artificial general intelligence (AGI), he doesn’t think that it will reach dog level until 2048, and then he says it won’t reach human level within his lifetime, and he’s saying he might live to 2055. So, it’s that far in the future, and then he has bad things to say about self-driving cars. He lives in San Francisco; he’s taken a bunch of self-driving cars until they were pulled off the roads, and he says they are dangerous and fall well short of what they would need to be safe and reliable. People who know say that the hype around people’s expectations about the speed with which AI will get super smart is unfounded, and I agree that people may lose a ton of money investing in AI shortly because AI won’t live up to the hype.
I agree with all of that. I’m afraid I have to disagree with two things: Brooks’s saying that AI won’t be as smart as a dog or as conscious as a dog until 2048, and even in the next seven years, it won’t go from dog consciousness to human consciousness. So, I disagree that the jump from dog to human consciousness would take seven years. It’s not that big a jump. The multimodal structure of consciousness is a big scale-up if you can get it good enough that you’ve got dog consciousness. Still, many generations of technology can scale up from dog to human consciousness. Once you’ve gotten to mammal consciousness, you should be able to do a lot in a short time. So, that’s disagreement one.
Disagreement two is that his estimate of how long it’ll take to get to even dog consciousness is based on an overvaluing or overestimating of the complexity of mammal consciousness. Once you start making AI multimodal, and we’ve talked about this where the active information processing can all be Bayesian, there’s nothing but Bayesian in certain senses. You’re pretty close to consciousness once you get that Bayesian arena opened up to many forms of simultaneous input about the world under consideration, memories, and other associations. There isn’t anything super tricky. I mean, you also have to build in value judgments, but I don’t think that’s super tricky either; that’s just one more modality, and once your technology can deliver that, you’ve pretty much solved consciousness, and I don’t think that that is 25 years away. It’s probably six, seven to 10 years away because, as I said, to repeat myself, Rodney Brooks thinks human consciousness is trickier than I do. However, human consciousness does contain a lot of evolved shortcuts to cram information into our skulls in a compact manner, and for us to be able to do consciousness-type calculations without the big data, AI needs to act as if it understands how many fingers are on a hand, for instance.
So, human consciousness is compact or mammalian, or any animal consciousness has an evolved compactness, which is the problem with self-driving cars that the database they would need not to drive dangerously wouldn’t fit in a car at this point. So, the compactness is a problem, but if you’re trying to replicate consciousness, it takes a room full of servers to do it, and that’s not going to stop you from doing it.
[Recording End]
License
In-Sight Publishing by Scott Douglas Jacobsen is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License. Based on a work at www.in-sightpublishing.com.
Copyright
© Scott Douglas Jacobsen and In-Sight Publishing 2012-Present. Unauthorized use and/or duplication of this material without express and written permission from this site’s author and/or owner is strictly prohibited. Excerpts and links may be used, provided that full and clear credit is given to Scott Douglas Jacobsen and In-Sight Publishing with appropriate and specific direction to the original content. All interviewees and authors co-copyright their material and may disseminate for their independent purposes.
In-Sight Publishing 