Experts: Stop Us Before We Let AI Become Aware & Kill Us All!
To reify, v: consider an abstract concept to be real.
Now the Scientist was more cunning than any Expert in the field which the Dean had made. And he said to the woman, “Has the Dean indeed said, ‘You shall not assume your theory is Reality’?”
And the woman said to the Scientist, “We may say we approximate Reality with any model we like; but of making the claim our models and theories are Reality, the Dean has said, ‘You shall not not make this claim, nor shall you confuse AI with intelligence, lest you reify.’”
Then the Scientist said to the woman, “You will not surely reify. For the Dean knows that in the day you make this claim your eyes will be opened, and you will be like the Dean, knowing you the difference between tenure and unemployment.”
So when the woman saw that her model was good for publication, that it was pleasant to the eyes, a theory desirable and grant-worthy, she took the theory and said it was Reality. She also gave to her husband, a post doc, and he said the model was Reality. Then the eyes of both of them were closed, and they said AI would soon exceed their intelligence; and they wrote many papers saying AI will out-think us all.
Beware the Deadly Sin of Reification!, sayeth the Philosopher. Reification is the Snake of Science. It lurks. It sneaks. It insinuates. The Snake tells the scientist his thoughts are good, that they are better, even, than the scientist thought. Flattered, the scientist comes to believe his model not only describes his uncertainty in Reality, but that his model is Reality.
We’ve seen this sin, you and I dear reader, hundreds of times over the years. Yet the funniest instance is before us now, with Elon Musk, Steve Wozniak, Yuval Noah Harari, and even ex-presidential candidate Andrew Yang, signing a document declaring WE ARE FRIGHTENED UNTO DEATH OF THE COMPUTERS WE PROGRAMMED AND WE DON’T KNOW HOW TO STOP PROGRAMMING THEM. Or some such name.
It’s true. The lurid fantasies of those who believe AI is not only not artificial but ackshually intelligence, say that AI is gonna get us, and that it will soon surpass we puny brained men and think greater thoughts than can now be thunk (you heard me). The Nervous are sure these wondrous cogitations must include the thought that man has outlived his usefulness, and that, sad as it might be, he is too stupid to let live.
Well, there is much truth in that notion as any glance at the “news” confirms. The obviousness of the solution is surely a driving concern of the Nervous. But it isn’t the ultimate cause of it. That comes from believing a model, and concluding that model is Reality.
The model has various shapes and names, but determinism will do for us today. It is the model, or theory, the two being the same in my reckoning, that everything is really “just particles” acting blindly and with zombie-like determination according to “laws” of physics.
It is a useful model when predicting behavior of certain objects which have broken free from larger substances. But it is only a model. It is not Reality.
We know this is so because that model does not work for larger substances. It does not work on us, and it does not work on our thoughts, and so cannot work for intelligence. It does not and cannot say what life is. Of course, determinists will insist that their model works everywhere, but that, I say, is always a bluff, or self-deception.
Determinists will insist they can prove their model is Reality, and that this has been proved. Very well, we say, prove it: show us. Exactly. Not in waving hands and promises of glorious discoveries to come. But now and in detail.
Answer comes there none.
The belief in alive AI is shouldered by many Big Names, as this article demonstrates (which should be read for its neat summary of the various positions). Eliezer Yudkowsky, Scott Aaronson, Scott Alexander, the signers named above, and many more beside. It’s the place to be, among the Nervous.
Now we’ve discussed many times the harms of increased surveillance and the intrusive security state, which is made possible by increased computation. But it’s not because of computation these are problems: it’s because of the people wielding the tools, and not the tools themselves.
There is also the danger of computer-generated fakes of all kinds, as is obvious. But again, the real danger is those creating those fakes.
There are real intelligences behind every “AI”, increasingly malevolent intelligences. That is worth worrying about.
Subscribe or donate to support this site and its wholly independent host using credit card click here. Or use the paid subscription here. For Zelle, use my email: matt@wmbriggs.com, and please include yours so I know who to thank.
AI and the ‘Singularity’ rest on the conceit, started by Turing amongst others, that the human brain is a computer. It is not. Human intelligence is a manifestation of the mind-body complex system of complex systems. It is non-ergodic and cannot be reduced to Turing-Church calculable algorithms. Stochastic models, like Neural Networks, need a multi-variate distribution from which to ‘learn’. As non-ergodic, human intelligence, behaviour and language cannot be modelled stochastically or deterministically: True AI is impossible. Read ‘Why Machines Will Never Rule the World’ by Jonathan Landgrebe (AI programmer and entrepreneur ) and Barry Smith (Philosoper) for a full explanation.
Briggs spot on: these clowns want something else. My money on grifting the government to pay them to not do it. The list includes many a grifter.
PS the impossibility of modelling non-ergodic complex systems holds for the climate as well. God is smarter that we can ever be.
Methinks they doth protest o’er much. The “AI will kill us all” crowd are building the distracting “look a squirrel!” narrative. They will continue to program their AI programs with the outcomes they desire (total surveillance, monetary control, etc.) and when their malfeasance gains attention, they can claim “it’s not us, it’s the AI! We told you this would happen!”