Berkeley's new $5 million grant, and an introduction to Artificial Intelligence
What is artificial intelligence?
Well it's partially the reason that you get so many targeted ads, but it's more than that!
All of these separate fields combine to form something greater than the sum of their parts - Artificial Intelligence
Now I'm sure most of you have been exposed to AI in some form or another, either through Spielberg's film of the same name, or Iron Man, etc. and you're saying, yeah Chandler, AI is cool and all, but I don't want SkyNet to take over my life! That's where Human-Compatible Artifical Intelligence comes in.
If you asked an "un-friendly" AI to make you as many paper clips as possible, after awhile it would realize that it could break you down to turn you into more beautiful, beautiful paper clips!
which we don't want.
Berkeley just received a $5,555,550 grant to prevent these potential existential risks that could result from advanced AI. The Center for Human-Compatible AI will be led by Professor Stuart Russell, co-author of Artificial Intelligence : A Modern Approach - one of the most widely used textbooks in the field.
Research will focus on developing and aligning AI values with those of our own (teaching JARVIS that turning us into paperclips would be bad), the properties of an ideal value system, the structure of our own value systems, and much more.
Thank you for reading!
Image Sources, in order of appearance:
source
I had to laff when google AI went all facist and started swearing, but its a bit worrying if they go all terminator on ya ass
lol true. It's all fun and games until someone gets turned into a paperclip!
no, none of those hueristic processes will ever rise to intelligence, even if you combine them all.
SIRI is not the path.
The path lies in unstructured neural net systems, or, rather, tiny decision processes with extreme connectivity. If a human can recognize the algorithm, then it's not the way.