Join John Adams, world renowned Intl Matchmaker, Monday nights 8:30 EST for Live Webcasts!
And check out Five Reasons why you should attend a FREE AFA Seminar! See locations and dates here.
View Active Topics View Your Posts Latest 100 Topics FAQ Topics Mobile Friendly Theme
Discuss and talk about any general topic.
4 posts • Page 1 of 1
Its here and for real,
https://www.youtube.com/results?search_ ... s+Hassabis
So far we have had what might be a kind of narrow AI, like Amazon's Echo, IBM's Watson or even self driving cars. But this is a fully working neural net, very different. Demis Hassabis has sort of created a synthetic hippocampus. The YouTubes can give the details. This has been tried since the 80's with no real success. Really only in the last few years have we had the computational power to do this. Another very telling thing they say, "it is surprisingly easy to make this work".
Its no joke this kind of thing is going to completely disrupt jobs, I think it will effect most all of them. Not to worry, there will be all kinds of new ones, but it will be a very different world. And the transition will be hard. The Luddite problem is often forefront, everyone speculates there is going to be massive unemployment. But what if the opposite becomes the actual reality, massive employment because of AI? From this point in time the post-modernist/SJW's/progressives are the biggest risk with there identity politics and collective guilt to those who do not completely agree with there one dimensional narrative.
This kind of tech fully matured would make a sex bot that no woman could compete with and that is not just physical. This still is several decades away at the minimum as the human brain would need to be fully reverse engineered. I think a sex bot in a decade, compared to today's sex bot, will be a very different experience if even the current neural net type of AI stuff is applied. If this were done this sex bot could fool some people for a while that she was actually human. In the far future the easiest way to tell a real human is to look for flaws.
This is some narrow AI players
Don't buy into the hype. CNN's are really useful for certain tasks but they are not scalable and you can't really achieve General AI with them. They're just a semi-brute force untrackable statistical gimmick. Look at the AlphaGo video, they 'trained' it with 100,000 preexisting games, that's hardly as intelligent and 'general' as Hassabis implies.
Even if directly mimicking the brain was the way to go, which is hardly arguable, current neural models are naive at best and still lack a lot of the parameters.
The best work on this as far as I've seen is from Jeff Hawikings
https://www.youtube.com/results?search_ ... f+hawkings
1)Too much of one thing defeats the purpose.
2)Everybody is full of it. What's your hypocrisy?
There's a few major problems with artificial intelligence. One is that it undermines public influence. A way to force people of means from trampling people with less is to "give their business a heart attack." Depriving them of labor can nullify an action being undertaken in itself or give leverage against the movements of those making them. The public can be a pack of assholes sometimes, but the effort to conquer it is usually an unworthy one.
A different one is the theory of putting someone's brain on a computer & altering it. Sure, there's almost definitely some "pitchable positives," like with pretty much everything- but the negatives are there & are extreme. Considering that someone truly perverse can always have their brains blown out, there's no sense in giving someone that might very well be that perverse control of minds in the direct sense.
Another one, also in the case of putting people's brains on computer, it could be used to torment someone virtually for, theoretically, eternity (or at least for MUCH longer than the person would live otherwise). This is something a lot of people wouldn't presume, but things might not go according to their tastes.