/sci/ - Science & Math

>>6639105
The singularity crowd largely based their argument on the overly simplistic notion of equivalent computing power, as in number of transistors, equalling "intelligence" as in number of neurons = AI of that level is magically created. This is bullshit, but on the other hand AI research has made some real advances recently with "deep learning". They're still basically fancy expert systems with no actual intelligence that can only do one thing, but their capabilities are impressive nonetheless. Microsoft research has one that can translate from English to Chinese speech in real-time despite not being programmed with any translation rules, it just has a huge corpus of equivalent texts that it builds a statistical inference database from. It's realistic to believe we'll see AIs that could consistently pass the Turing test within a couple of decades, though true intelligence in the "hard AI" sense is a much harder problem given how little we still really understand about cognition or large-scale brain function.

>>6545481
Here's a must-read Asimov short story on the topic:
http://www.multivax.com/last_question.html

Follow that up with some good hard SF for serious explorations of this, maybe Iain M. Banks "Culture" books. However there's a pretty good chance we'll wipe ourselves out first, or just get taken out by a huge natural disaster before we get offworld.

For a long time I was sceptical about the singularity (it's still oversold by its advocates) because having as many transistors as neurons in animal X in no way automatically translates to equivalent intelligence, and hard AI research had stagnated for decades. However recent advances in Deep Learning are genuinely exciting, e.g. Microsoft recently built an AI that could near-realtime translate from English to Chinese with high accuracy *despite not being programmed to do so* just by absorbing an enormous corpus in both languages then doing statistical inference to guess meanings. This is an alien way of "thinking" from our perspective and will lead to AIs that work in ways we don't understand, especially when they start designing their own progeny, but it certainly seems like the way to go.

https://www.youtube.com/watch?v=Nu-nlQqFCKg
( The actual translation demo bit starts around https://www.youtube.com/watch?v=Nu-nlQqFCKg#t=7m10s )

>>6545942
Yeah "Moore's law" is unfortunately named since it was never a law, more like "Moore's observation of an interesting trend that has held for a while."

>>4646370
>>4646298

>Transhumanists
Pic related

>>3854119

What bothers you about math/sci notation? Personally sin²(x) has always seemed like an arbitrary stylistic decision. sin(x)², right? it IS sin(x)*sin(x), right? Arbitrary.

And using the same characters for a ton of concepts. And don't say variables don't matter because proof you're wrong: replace theta, n, x, y, etc with 1, 2, 3, etc and measure the change in your productivity when working problems. Function and form are fucking related, that's entry level revelation.

inb4 it is a flawless system that is not meant to change ever.

You. Yes, you. The ones who think science will save you, make you immortal, etc., quit living in a fantasy world. Just because your generation understands computers better than the last does not imply you will become the elders of the human race.

You will die.

That is all.

>transhumanism

Please support this petition to ban discussion of Transhumanism from /tg/ and /sci/, Transhumanism seeks to take everything good about humanity and replace it with cold, unfeeling machines.

Fapping to dresden codak in a transhumanist world would be the norm.

Movie was better.

also, saging futuristic blah blah blah and wishful thinking

HAHAHAHA

Life sucks and will always suck. It would be better if everybody just killed themselves now

As before....

>>2015179

W/E
bunch of priviledge 14 year old talking amongst themselves

jerk offs

2/2