When will we achieve human-level AI?
>>7899875
When our alien overlords who run the simulation we're in give us higher IQ so we can build those AIs.
>>7899875
between 2025-2035
>>7899940
lmao
what makes you think that soon
>>7899940
I agree. You'll start a shitstorm by posting that on /sci/, though. The anti-Kurzweil brigade should be here shortly...
>>7899982
Not him, but I also agree and I don't give any credence to what Kurzweil says. He'll just happen to be right this time.
>>7899875
we'll likely go extinct before we get anywhere close to true AI
>>7899976
The dumbest human plays atari games worse than deepminds AI, so we're already beyond low-end humans equivalency.
>>7900180
No, deepmind's AI only plays specific games well.
Take a look at the chart at the bottom of this paper that shows DQN scores versus human scores: https://www.cs.toronto.edu/~vmnih/docs/dqn.pdf
It's terrible at q-bert, seaquest, and space invaders.
>>7899989
How dare you question him. The unfaithful shall be left behind in the coming singularity.
There he is. There he goes again. Look, everyone! He posted it once again! Isn't he just the funniest guy around?! Oh my God.
I can almost see your pathetic overweight frame glowing in the dark, lit by your computer screen which is the only source of light in your room, giggling like a like girl as you once again type your little AI thread up and fill in the captcha. Or maybe you don't even fill in the captcha. Maybe you're such a disgusting NEET that you actually paid for a 4chan pass, so you just choose the picture. Oh, and we all know the picture. The "epic" Robots are gonna take over meme, isn't it? I imagine you little shit laughing so hard as you click it that you drop your Doritos on the floor, but it's ok, your mother will clean it up in the morning. Oh, that's right. Did I fail to mention? You live with your mother. You are a fat fucking fuckup, she's probably so sick of you already. So sick of having to do everything for you all goddamn day, every day, for a grown man who spends all his time on 4chan posting about a AI book. Just imagine this. She had you, and then she thought you were gonna be a scientist or an astronaut or something grand, and then you became a NEET. A pathetic AIfag NEET. She probably cries herself to sleep everyday thinking about how bad it is and how she wishes she could just disappear. She can't even try to talk with you because all you say is "MUH TECHNOLOGICAL SINGULARITY." You've become a parody of your own self. And that's all you are. A sad little man laughing in the dark by himself as he prepares to indulge in the same old dance that he's done a million times now. And that's all you'll ever be.
Forever...
Never
>>7900227
Yes, that's their old thing.
They can now play every single one of them at superhuman level, and are tackling quake and starcraft.
Ain't been peer reviewed yet but you can find it out if you follow on twitter people that gets invited at conferences where Demis Hassabis/David Silver speak
>>7899875
'we' wont.
When, IF, this ever happens, a tech will have to have been invented that could bring itself to said level of intelligence.
>>7900457
you're a shit butt.
>>7901601
>a tech will have to have been invented that could bring itself to said level of intelligence.
We already have various automated learning/evolution algorithms.
Your idea of AI seem to come entirely from movies and memes.
>>7899940
There's no fucking way to predict that.
We can be sure it'll be more than 10 years, 20+ realistically, but you need to define "human-level".
It could be 20 years, it could be 50 or 80 years.
We just cannot know.
But special purpose AI used to replace the vast majority of jobs, that'll come much faster, possibly in 20-30 years we'll see a huge increase in such systems.
AI doesnt have emotion so of course it will skyrocket and surpass human intelligence in the near future
>>7899875 (OP)
We already did.
>>7901596
>AI can play quake better than humans
It's a first person shooter, of course a computer is better at it than a human, their response time is <1ms and they don't miss.
The way we are headed we are gonna have fusion power and warp drive before we make AI. Kind of like Star Trek where Data was so hard to make.
>>7903712
I doubt it, really.
I can talk to my phone and ask it to call "X business name" and it will just do it. Weird example, but Google Photos Assistant learned what a fursuit is and can detect photos of the same person from childhood to adulthood.
The field of machine learning and AI is accelerating pretty quickly. I mean, think of the AI available to consumers 10 years ago and the AI available to consumers today.
>>7900180
Yeah, but the thing is, that AI will never design games that sell as much as Atari Games.
>>7903759
>never
How long are we talking here? 100 years? Thousand? Million? Googleplex? Graham's number? Countable infinity? Uncountable infinity?
>>7903452
So it will not be an AI. Psyche, the awareness, the consciousness rises from and only from emotion, the experience. There is no intelligence without experience.
Have a thought, what the cognition, the thinking, really is? Isn.t it just complex of different emotions? Just like every perceptional data, the verbal monologue you have "inside your head", is complexes of emotions, aren't they?
>>7903764
The last one.
>>7903777
Then you are limited indeed
a program is generally faster than a human if it doesnt learn or anything. if it just has to execute a function it is really effective.
but to really be an ai it has to first find a function to execute. now you have to include the time it takes to train your ai in the time it needs to solve problems.
if you really want an AI that is human like, you have to stack layer of layers of learning algorithms.
it needs to learn how to learn how to learn how to play a specific game.
we humans do this since our birth. compared to an attempt at an AI we have an advantage of years of training.
Ive started to think that AI wont easily be smarter than a human...
>>7903561
Funny how something don't count as a metric for AI as soon as AI can do it better than humans.
>>7903788
>we humans do this since our birth
>Ive started to think that AI wont easily be smarter than a human...
You just keep in mind that nature has been testing these learning algorithms for around 3,5 billion years. It tried numerous platforms from incredibly small to incredibly large to everything in between. We've been going at it for like a few decades...
>>7903774
You are applying a definition to Artificial Intelligence which doesn't necessarily exist.
Please watch the following videos in order:
https://www.youtube.com/watch?v=q6iqI2GIllI
https://www.youtube.com/watch?v=tcdVC4e6EV4
Presenter has a PhD in Computer Science. They will help you to understand that you are anthropomorphising heavily. Consciousness doesn't have to resemble human consciousness or even general animal consciousness.
>>7903800
>Presenter has a PhD in Computer Science.
i would mention that this is an obvious appeal to authority, but that's not much of an authority at all is it. toplel.
>>7903803
>appeal to authority
Top kek.
What I actually did: provided evidence for a claim and described the source's level of knowledge on the topic.
Please keep your Philosophy I terminology to yourself when adults are speaking, especially when you do not have a firm grasp of the terms you are using.
>Human level AI
Fucking conspiracy/comp sci
