Are we close to building the first artificial consciousness?

>>56121611
watson is pretty close

as far as consumer AI goes we're still stuck with shit like cortana and siri

Short answer: No

Long answer: How would we know?

>>56121611
It'd be pseudo-consciousness. Full-fledged consciousness is achieved only through brain-like processing units, like the NPU. Since we mainly code our applications and illusio-intelligent systems on computer architectures that look and function differently from the biological brain we may see a true artificial sentience being developed in the far future.

the closest we can get to is growing human brains with sensors that can read neurons and translate those signals into computer code for output. The rest is up to using pnumatics or batteries to power the artificial body.

Two things are needed:
1. a material cable that can attach the brain to a computer chip without infections. probably microscopic and synthetic enough to withstand long term useage.

2. a small powerhouse that has alot of output from minor input to help regulate the brain and the rest of the body.

Nope. We don't even understand what consciousness really is and how it works.

>>56121956
>>56122256
But the issue with doing this is that you're not actually improving on a person. If the best you can do is 1:1 replication of how a human mind functions you've achieved nothing beyond a really complex and expensive form of fucking without any of the fun.

>>56122401
if you would do a 1:1 replica you would still have a normal person with near instant access to the nearly all human knowledge

>>56122450
Give a person a smartphone. Boom. Done. Unless you can make a person that can interface with this in a useful way you haven't made anything helpful.

>>56122461
anon, i...

People very falsely think that human intelligence is somehow higher than computer intelligence. The amount of memory a human can hold in their head at one time is pitiful, the amount of information in memory in relation to sensory input that a human receives is extremely pitiful. The simplest computer never forgets. When quantum computers come online with a significant amount of qubits the processing power of a computer will outstrip anything a programmer could put in to a computer and computers are going to start solving math and physics problems that humans never even pondered.

People seem to think that awareness or self awareness is some kind of metric to computing complexity and its not. Awareness is just a memory loop. A memory loop is a very basic thing that the tiniest microbe possesses.

>>56122497
TIL microbes are self-aware /s

https://en.wikipedia.org/wiki/G%C3%B6del%27s_incompleteness_theorems

on another note, I have personal reservations, that, human consciousness, or even any living consciousness for that matter has it's roots deeper than simple, electical pulses and the "physical" universe, not to jumping to the spiritual explanation, lets skip that for something tangible, even the electron, we know little about it, and most people know less than the previously implied "we". it can reach well into the spectrum of spiritualism I suppose, concepts of the self outside the self, etc. but that's another topic.

"can we make a robot into a hum"
no. no. no. n-NO. why?
maybe this generation is spoiled with technology, especially the kids who didn't grow up with pogs or ham radios. your super sexy robot is essentially bonzi buddy. If-Thans, even if you scripted a new language, it is a physical extension of the coders and engineers.

AI could only ever mimic being human, even if in some subjective sense it could emulate human emotions, even though it doesn't have the evolutionary, biological, cellular, neuro-structural to form them, notwithstanding education.

words don't do it justice because psychoanalyzing AI that doesn't exist, well it's absurd.

point is, lesson learned from that film ex machina is, don't be the tool's tool, you tool. and don't delude yourself that it isn't the case when it is.

To create an artificial consciousness we would have to first make a computer that is flawed by standards of computer science. A brain functions fundamentally differently from a computer and therefore is an unattainable goal in the near future from the stand point of modern information technology. Our computers can do accurate and complex physics simulations in a matter of seconds, but a brain can approximate even more complex simulations in a 100th of the time.

Heavens, no. We can't even solidly define consciousness, much less engineer it.

AI techniques are now getting very good at a few specific concrete tasks involving recognition and response. Self-esteem cognition, however, is so far off I couldn't even give a figure in decades. We don't even know where to start.

>>56121611
Source on the pic op?

I'm sure most of the comments sum this up but no. We know _very_ little about how our own brains work. And this is mainly because we can't see how a brain properly works without it being inside a host. Unless some sort of miraculous breakthrough happens that allows us to fully understand our own brains, it's likely that this won't happen in our lifetimes.

The closest to an AI we have accomplished has already been around for decades: machine learning. Then again machine learning is loosely based on how we learn and make decisions through reinforcement. It's in no way a 'replication' of how our own brain actually works.

>>56122664
Thanks Hofstadter but your wife is still dead

why does Elon Musk, Stephen Hawking and Raymond Kurzweilbelieve that consciousness will come to AI by 2030?

>>56122401
>>56122450
https://www.qualcomm.com/news/onq/2013/10/10/introducing-qualcomm-zeroth-processors-brain-inspired-computing

>>56121611
i wanna know if there is anything on this level yet https://www.youtube.com/watch?v=qDJRf5UVwLY

I'm pretty sure it's impossible to create consciousness artificially.

File: 015.jpg (624KB, 1280x1824px) Image search: [Google]

624KB, 1280x1824px

>>56121611
>>56122955
Not OP, but the source is...

>Osake ni Tayoranakya Sex no Hitotsu mo Manzoku ni Dekinai. _ When I'm Drunk, I Might Be Able To Have Sex With You
http://g.e-hentai.org/g/933694/809bfdca70/

>>56123633
I think it's possible, you just need the proper technology to create a organism complex enough.

>>56121611
Techies believing that achieving consciousness in machines simply is a matter of time have a poor understanding, not of computation, but of the nature of the problem itself. Trying to take consciousness into consideration when formulating some third-person theory of mind is a problem in itself, let alone simulating or reproducing it.
http://plato.stanford.edu/entries/consciousness/
>>56123057
Underrated

>>56121611
First you need to define conciousness.
If conciousness is simple memory loop like other anon said then yes we can do it. It depends on definition

Yes

t. Android

>>56124001
In order to understand how consciousness works first we need to understand how life works, because only living things can be be self-aware.

>>56121611
Fuck off hetshit and take your trash with you.

>>56121611
I've heard somebody succeeded, I can't give a source though.
But from what I also heard, even though it may be incredibly useful, it's still too risky to use, since there can be no connection to the outside from the computer that this thing is running on.

>>56121956
>Full-fledged consciousness is achieved only through brain-like processing units, like the NPU.

That's pulled out of your ass. Literally no-one knows what causes consciousness and how or even whether it would be possible to reproduce it.

>>56122497
You falsely think that speed correlates directly to inteligence. An elephant can remember stuff for a longer time than humans, are they smarter?
Computer inteligence is extremely simple, you can break it down to simple assembly commands. Also, self-awareness was never a real measurement of intelligence, i can make a computer say that it is self-aware, how will you know it isn't?

>>56121611
Human emotions are nothing more than a balance of hormones and other chemical components.

>>56121956

in ~16 years, assuming trends continue, we will be able to run deep neural networks with as many connections as a human brain

>>56122401

electrical connections are faster than biological neurons so an equivalently connected net would in principal be more powerful than the biological equivalent

see work on deep nns by Schmidhuber

>>56122664
kek

>>56123068
>why does Elon Musk, Stephen Hawking and Raymond Kurzweilbelieve that consciousness will come to AI by 2030?

None of those people are involved in AI/machine learning research. Anyone who knows anything about it will tell you we are many decades away. The state of the art now are various techniques based on deep neural networks. Convolutional neural networks for image recognition are probably the ones you've heard most about.

These are just really fancy regression/function estimation algorithms. They all train through gradient descent with backpropagation, which is an algorithm that would not be applicable to an actual AI.

People talk about Watson and shit. Those are expert systems that just apply rules. It is not applicable to general AI.

Convolutional neural networks can achieve some generalizing (like seeing certain features on a dog, and then concluding that another animal with those same features is a dog). But they need thousands of times more examples than the human brain does.

The generalizing performance of the brain is incredible, light years beyond what ImageNet can do. I don't know when AI will come out, but I do know the hardware will not resemble anything currently used, and the software will have basically nothing to do with the AI technologies we have today.

File: maxresdefault.jpg (34KB, 1280x720px) Image search: [Google]

34KB, 1280x720px

>implying you are not a robot programmed to think that you are an organic being

>>56122497
>awareness is just a memory loop
Wow, did you never learn how to bullshit in high school? You just blather on like a retard

>>56126036
>electrical connections are faster than biological neurons so an equivalently connected net would in principal be more powerful than the biological equivalent

"Neurons" in neural networks /= actual neurons.

An actual neuron is a complex machine that no one even really understands yet. This is a ReLU neuron from a neural network:

max(0,x); If x is positive, it's x, if it's negative, it's zero.

Not really comparable. Also, practical neural networks have a very specific architecture, where the neurons are arranged in layers, and neurons only connect with other neurons in layers directly above or below. This is very restrictive, but it is intractable to train other, more general architectures.

Not only that, but standard neural networks have dense connections between layers (every neuron in one layers is connected to every neuron in the next). This is not computationally feasible as you scale up the numbers.

The obvious thing to do is to not have every neuron connect to every other one. This is what the brain does. However, this doesn't work for artificial neural networks in general, because sparse connections like that can't be trained with gradient descent. The best you can do is something like simulated annealing/MCMC, which is literally thousands of times slower.

>>56125082
The NPU functions exactly like a biological brain. Theoretically achieving consciousness with its architecture is possible.

>>56128460
So you're saying a silicon processor functions exactly like a biological brain. How retarded are you?

Also, perhaps you shouldn't be going to Qualcomm's website for information on this. They are the ones trying to market this shit, so they will make it sound more revolutionary that it actually is.