I've heard a number of arguments back and forth as to why we havn't achieved our vision of AGI,
'wrong approach', plausible - there are many possible approaches to AI machine learning, each with their strengths and weaknesses, and many have only just began to be explored.
'not enough processing power' almost certain to be wrong, if not in the present, soon. Certain approaches, such as simulating a human brain, require far more processing power than we currently have access to.
However i'd wager that algorithmic approaches to the brains processes would take a fraction of processing power than what it'd take to brute force the problem.
'The human mind contains something that cannot be simulated on a turing machine' ...GTFO.
Additionally, processors are complex, more cores on a chip is, as you said a workaround to limitations to our silicone based processors. Its not just a matter of the problems with quantum effects either, its also about our ability to cool them (I remember something about chips hotter than the surface of the sun...)
From what I understand most of a 'core' is dedicated to crazy optimizations (out of order execution, branch prediction, pipe-lining, hyper-threading etc.) rather than 'simply' executing instructions. And I hardly doubt new innovations will stop along those lines anytime soon.
Moving into the foreseeable future, parallel computation is where it is at.
Addendum: The main problems I see in the future are social and legal problems, most technical problems will be addressed in due time.
How does one sue a machine if something goes wrong? How do we cope with AI, and the current 'planned obsolesce' movement? Do machines, intelligent in their own right, get rights - or are they merely 'tools that talk'? What rights do they have over their own source code?
And how does one compare an actual, right holding artificial citizen with a simple utility drone? Are there varying levels of rights (and responsibility)for levels of intelligence? Should the rights and responsibilities of (some(?)) AI go above and beyond our own?
These questions, and many more, coming up in YOUR future, stay tuned.
TL DR: The future is deep, mannn.
Goto original forum post.
Saturday, November 9, 2013
Friday, January 11, 2013
0.o Shiney
So for a while now i've been slowly reading up on language, it started with the word nerds (link is to the older archive, there is a newer set out there too)
This continued with drilling down down through the corridors of linguistics to base levels where there is some rather cool things; For example I find it oddly entertaining to listen to someone with a great deal of verbal dexterity vocalize letters of the International phonetic alphabet.
The highlight reel includes the fact that I have found out that SAPI (microsoft's TTS/Speech recognition... thing) while powerful, has a long way to go, and that there are some rather interesting papers on alternate means of speech recognition out there.
Expect more articles to do with speech and sound - particularly along the recognition side of things. TTFN.
TL, DR;
WORD WORDS WORDS WORDS WORDS OH GOD THE WORDS also sounds that makes up MORE WORDS.
This continued with drilling down down through the corridors of linguistics to base levels where there is some rather cool things; For example I find it oddly entertaining to listen to someone with a great deal of verbal dexterity vocalize letters of the International phonetic alphabet.
The highlight reel includes the fact that I have found out that SAPI (microsoft's TTS/Speech recognition... thing) while powerful, has a long way to go, and that there are some rather interesting papers on alternate means of speech recognition out there.
Expect more articles to do with speech and sound - particularly along the recognition side of things. TTFN.
TL, DR;
WORD WORDS WORDS WORDS WORDS OH GOD THE WORDS also sounds that makes up MORE WORDS.
Tuesday, January 8, 2013
Chip'n Review - Robots/Talking Robots Podcasts
Robots podcast - Four and a bit Chips
Robots podcast -and by extension- the discontinued Talking Robots podcast are a pair of great publications about, you guessed it robots.Why am I lumping them together?
- Roughly the same people involved effectively
- Same continuity (one started right after they discontinued the other).
- Similar content
So what do (these) podcasts offer?
In a nutshell Cool technology, awesome (and insightful) interviews, top notch production quality and if you squint really hard insights into -wait for it - THE WORLD OF TOMORROW!
These podcasts are by no means technical meaning anybody with an interest in robotics can listen to episodes and since they generally sit at around 30 minutes long they don't outstay their welcome.
TL DR;
May include traces of anything remotely robotic, medium length, no big words inside, you like - but only if you're a fan of such.
Subscribe to:
Posts (Atom)