Artificial intelligence still has some way to go

"idiots like Musk" c'mon.

Also, this is interesting:
http://www.independent.co.uk/news/science/why-evolution-may-be-smarter-than-we-thought-a6839186.html

i mentioned gates/hawking/woz/musk because they're more well known, but the letter they signed concerning autonomous weapons was also signed by hundreds of leading AI researchers.

i guess i don't fight back often enough (here or IRL), and i often shoot myself in the foot by talking shit on myself before others can, but to reduce the warnings of a bunch of leading researchers in the field as "smart celebrities" is kind of baloney

there's no way to prove that they're right or wrong - it's speculation about danger many decades away. it's pointless. so i'm not exactly tying my ego to the outcome of what a bunch of people think about this. but i do get incredibly annoyed by people who feign certainty about something that it is impossible to be certain about

Musk does some good stuff (I am all for SpaceX) but then also comes up with and says a lot of dumb shit (Hyperloop) so yeah I don't have a ton of respect for him

Autonomous weapons are scary! I think everyone agrees with that. Superintelligences turning the world into a massive typewriter (or w/e the example is) is maybe less of an imminent danger.

not everyone

Autonomous Weapons

I'm no celebrity, but I'm pretty convinced that some deep, powerful AI is around the corner. A lot of money is being poured into this right now, and not just into brute-force type stuff (which, at its best, might be good enough to be mistaken for AI, but does seem, intuitively, to lack consciousness), but also into systems that model neural networks and more opaque systems (evolving FPGA systems, memristor-based circuits, massively-parallel machine-learning things) that, to me, seem likely to actually generate something more generally intelligent or even conscious..

Dismissing Google's search algorithm seems a bit hasty, considering how powerful it is. No, I wouldn't ever call it conscious, but (I think) it is definitely intelligent.

in a large way navigating social media has made The Turing Test into an everyday banality

Google's search algorithm is so smart that it allows people to follow links to sites infested with malicious code literally all the time.
All the big technology companies in the world fight an endless battle against common criminals on a daily basis and haven't come up with a way to clear the web of malware being delivered via their own advertising networks. Do you understand why I don't have any confidence that any kind of impressive, stable AI is "around the corner?"

idk you're assuming google really cares whether you get malware

might be a selling point for their users

but I wouldn't know since I don't use Google lol

So now the bar is "smarter than teams of malware developers?"

I'm just saying that it's hard for me not to be impressed when I type in a couple of words, and within 8ms Google returns the page/video/news article I was looking for.

Okay, so you're a rocket scientist
That don't impress me, Musk
So you got the brain but have you got the touch
Don't get me wrong, yeah I think SpaceX's alright
But that won't keep me warm in the middle of the night
That don't impress me, Musk

lol

otm

i'm going to read this thread properly because i strongly suspect you are all wrong

but andrew ng's 'worrying about evil AI is like worrying about overpopulation on mars' is i think i useful way of thinking about how productive this debate is right now

(although i heard that at NIPS last month he changed this to 'worrying about overpopulation on alpha centauri')

glad to see the string theory experts posting here

i openly will attest to being wrong

ng otm

string.h theory experts

http://www.defenseone.com/technology/2016/01/refugee-or-terrorist-ibm-thinks-its-software-has-answer/125484/

Borene was careful to indicate that the hypothetical score was not an absolute indicator of guilt or innocence. “It’s like a credit score."

http://cs.jhu.edu/~jason/tutorials/ml-simplex.png

yep m/l

caek who do you work for that you write reports on probabilistic programming etc (if you can say)

Can't say but it's fun. Cnns, rnns, prob programming, all that good stuff.

Trump campaign then?

ha no that's this genius actually https://twitter.com/witolddc

http://www.fastcompany.com/3055702/data-pros-doubt-trumps-ground-game-in-iowa-will-succeed-if-cult-of-personality-fails

Yoshua Bengio talking sense:

The thing I’m more worried about, in a foreseeable future, is not computers taking over the world. I’m more worried about misuse of AI. Things like bad military uses, manipulating people through really smart advertising; also, the social impact, like many people losing their jobs. Society needs to get together and come up with a collective response, and not leave it to the law of the jungle to sort things out.

http://www.technologyreview.com/qa/546301/will-machines-eliminate-us/

yup.

i tell anyone who will listen to read section A of this article http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2477899

this is so much more than killer robots, and it's happening right now

no way

Advocates of algorithmic techniques like data mining argue that they eliminate human biases from the decision-making process.

oh come on anyone who does that sucks, this thesis is a given

"technology is a social phenomenon and it might have not-good consequences." kind of lightweight and myopic given that sts and actor-network theory have been kicking around since the 70s... xp

lol yeah anyone trying to claim technology has nothing to do with morals and is a pure science, as if it's a natural phenomena and descended from pure logic, they are the ones to be wary of

yes and these ppl exist is my point

i mean apologies if the claims in that paper are self evident to you, but they're not self-evident to 90% of data scientists

fair

Google’s AI beats world Go champion in first of five matches

Failing the Third Machine Age

^from 2014, btw, so possibly already consumed by some

Make a list of shit nobody actually believes and then cherry pick quotes by a wide range of credentialed and non-credentialed so-called smart people and hey look a listicle!!! my head
http://gizmodo.com/everything-you-know-about-artificial-intelligence-is-wr-1764020220

Related, and way more fun: http://www.clickhole.com/clickventure/youre-computer-can-you-pass-turing-test-4009#27,

ugh, the structure of that gizmodo thing, with "Myth" set against "Reality", only by "reality" i suppose they mean "fact", which they then support with...opinions? Myth: Someone thinks something might happen in the future. Reality: someone else thinks that it might not.

wtf

also they forgot to edit it. like near the end, with this quote:

“Over the next couple of decades AI is going to destroy many jobs, but this is a good thing,” Miller told Gizmodo.

who is Miller? they never say. it is a mystery

perhaps the author was distressed because his job was being destroyed as he wrote the article. and maybe the AI wrote the bit about it being a good thing.

sorry, but let me lightly edit to fit the format:

MYTH: perhaps the author was distressed because his job was being destroyed as he wrote the article

REALITY: maybe the AI wrote the bit about it being a good thing

computers can play go now, I guess

truly the end is near

people are learning new strategies from computers, in a 2500 year old game. sounds like the beginning to me!

look at these robots falling over!

https://www.youtube.com/watch?v=g0TaYhjpOfo

look at these stupid babies!

https://www.youtube.com/watch?v=u9nWOaDit5w

Those robots have far less sophisticated sensors and control mechanisms than the babies do. A baby can learn to walk without constantly falling over in a matter of a few weeks, without being torn down, re-designed, refitted or rebuilt. But the robots will probably need another 20 years, endless redesigns and hundreds of millions of dollars in research funding before they can open doors or climb over obstacles.

It seems to me both perfectly understandable and mostly unjustifiable to try to make robots replicate human form with human physical capabilities. We already have humans for that. AI is always more successful when it focusses on narrowly specific jobs and builds on that capability by increments.