Zuckerberg, Musk and Fake Steve Jobs invest in secretive AI firm Elon Musk and Mark Zuckerberg have become some of the latest investors in secretive artificial intelligence software firm Vicarious. The SpaceX billionaire and the Facebook founder took part in a $40m funding round, the second major cash-raising from the firm in the last two years, the Wall Street Journal reported. The actor …
You mean they are going to create programmes that spend their time downloading p*rn, (and kitten videos), play flappy bird, and alternatively write articles or inane comments on the Register?
Hmmm and how is that an improvement? We've got a few billion of those already !
Innovate Please....
I remember when Wolfram Alpha was going to do this. Instead it is answering Siri queries using the same definitely-not-AI search-based technology that Google uses for Google Now queries.
Every new AI startup claims they're going to simulate the brain in software, but find that since we don't actually know how the brain works, that doing so isn't so easy, so they fall back on writing software that uses methods completely different from what the human brain uses.
I don't think we'll ever achieve AI by simulating the human brain, because we don't even know exactly what, if any, role that quantum mechanics takes in the function of a single neuron, let alone across billions of them. We may someday have machines smarter than people, but their thinking will be as different from ours as the way a car moves from place to place is different to the way a human moves.
If they do succeed it will no longer be AI.
Despite 60+ years of failure people persist in thinking that the "obvious" way to duplicate a very large collection (10^10) of low speed (but very high fan out / fan in) elements operating in parallel is with a small number of high speed serial executing machines.
So why should this version be any more successful than any of the previous?
Despite 60+ years of failure people persist in thinking that the "obvious" way to duplicate a very large collection (10^10) of low speed (but very high fan out / fan in) elements operating in parallel is with a small number of high speed serial executing machines.
Because while those serial machines are indeed a poor direct fit for the problem, they also happen to be know to be able to simulate any information processing system given enough time, and they are also fortuitously easily accessible commercially. Once we have more than a faint clue of what we are actually trying to copy we might prefer to go the GPU way and create custom ASICs for the job which will undoubtedly perform closer to human real-time - but as long we are still throwing around theories, the current serial machines we have handy should / will have to do just fine.
Is a debug log, the things you are thinking about right now have already happened, you are finished with each thought before you become aware of it.
Human Intelligence isn't special, it's certainly not magic, it's just a shitload of algorithms running in parallel with a human readable log. The current AI efforts aren't failing because they haven't replicated 10^10 neurones, that's like saying that transistors can't replicate the valve based monsters that preceded them, they're failing because the mix of algorithms is really fucking complicated.
We're taking tiny steps towards AI and it can seem like it's never going to happen, but (prediction), it will occur suddenly, in a single great jump from barely started, simply because someone changed a parameter or made a minor tweak to an algorithm. Machine AI is inevitable.
/side note: In my youth I worked in print, I remember a manual we printed for a 'Red Baron' WWI flying ace game which contained background info on the main protagonists, the aformentioned Red Baron, an English chap I can't recall, and an American, Eddie Rickenbacker, born 1890, died 1973. Eddie was born before powered flight and died after man had stepped on the moon. I was blown away by the progress this man had seen in his lifetime, and for a long time it's seemed like it was an aberration, fuelled by three wars, two hot and one cold, but I'm starting to think the changes wrought in my lifetime will blow away a mere trip to the moon.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
