Have you solved AI yet?
The problem for AI is that it requires a plausible model of human cognition. Having some high-dimensional projection algorithms that are called "machine learning" is not a plausible model of human cognition.
>Deep Learning
Another fancy buzzword to join "data mining", "big data", "cloud", "enterprise quality software" etc.
>>55511972
but what about "muh realistic bedrooms"
>>55512070
It's called "Image Processing", no need for DERP LEARNING here also
deep learning = using deep neural networks = using any neural network with more than 1 hidden layer
That's literally all there is to that
>>55511972
>>55512161
>le buzzword
It dramatically lowers the training time. Try making a good image recognition program with classical techniques, faggot.
>just throw more le perceptrons at it!111
Fucking reddit.
>>55512535
I don't know why you're talking about reddit, but let me just add that I hate this piece of shit karma whoring retarded alien intro to CS 101 web dev tier website that requires registration and is crap.
>>55512535
>Try making a good image recognition program
why should I?
>>55511941
this tbhfam
>>55511941
This anon knows. Right now we aren't able to program a learning model capable of making accurate and independent extrapolations that could be considered "intelligent"
We can program machines that interpolate well and improve their efficiency within their pre-ordained limits, but if you have to have pooloo interns to do data entry for your 'A.I.' it defeats the point.
>>55511972
>Too stupid to understand that all of those terms have precise, defined meanings
/g/ - Computer Illiteracy
>>55512884
Is it still debatable that this is even possible? I read an article that claimed machines could never be truly intelligent and the gist of it was that they fundamentally cannot "read between the lines", or they are completely deterministic.
I suppose if a model of the human brain were fully developed it may show that we are deterministic as well though. I also believe that a machine powerful enough could emulate a human brain well enough for it not to matter if it is technically operating on the same fundamental level.
AI noob so maybe I am way way off
>>55512916
I understand that they have a meaning, but they're still buzzwords used for marketing.
>>55511941
>>55512884
>>55512833
The Chinese Room Argument. Even if, with the current or future ML algos, we somehow manage to train a computer to generate intelligent responses on a certain input, this does not mean the computer truly understands what it "knows"
>>55513103
The same logic can be applied to you, you fucking p-zombie.
>>55513186
>you fucking p-zombie
>same logic can be applied to you
H-how did you know? This is impossible, I beat all the turing tests!
It's true though, maybe my brain is just one fancy algorithm that spews feellings and thoughts based on a pre-determined set of rules known as "common sense". Or maybe I'm just a frontend, and all the real processing occurs somewhere, like pajeet's webapps. We will only know what we truly are once we figure out how the brain works and how, from this functionality, arise phenomena like conscience and intelligence.
GUYS WHAT IF GUYS
GUYS
WHAT IF WE CONNECT A HUMAN
TO A COMPUTER?
>>55513994
thank you so much for that post it's so perfect, if not poetical
*EDIT: i didn'expect so many upvotes XD
**EDIT: OMG someone gave me gold! Thank you gentlesir.
>>55512927
Well, to expand on my original point, what we'd really need is a theory of causation that isn't dependent on ceteris paribus. That is, people say "correlation is not causation" without having to define exactly what causation means. Since scientific theories are meant to be causal accounts of some phenomena, science will always be bounded by what it means for A to cause B. That issue is why science has made essentially no progress on complex systems problems, but instead runs forever on the buzzword treadmill. As in, "surely this buzzword will make headway on complex systems."
That's the issue that stops us from having a plausible model of cognition, it's simply not possible for science-as-it-actually-exists to make such a thing. Which is why we sub in all these shitshows like game theory and behaviorism and machine learning.
>>55513771
The actual difference between you and a computer is that you're a perceiving subject. That is, you're aware of yourself and your own existence, which is a property that computers lack. Because of this, computers are not "similar to" minds in any way that matters.
>>55513103
That's getting more into philosophy than actual programming and mathematics. If an A.I. only superficially understands the information it processes - only the patterns in the data, with no metacognition - it's still going to be a good enough system to interface with most humans.
>>55515034
True, but I was talking about not being able achieve strong AI with the current trends of ML.
>>55514939
Simulating the biological process of perception of the self and the environment would make a computer pretty similar to us, I would think.
