Agreed. It was a great post. I tend to think of my self as a "liberal-tarian." With the tenet being to set a goal for a standard of living and social / environment targets that benefits as many as possible, but anything beyond that get out of the way.
In this way I am for socialized health care, significantly stricter regulations or punitive measures regarding the environment. However for most other issues I would rather have a competitive market.
In the end I think some form of UBI may help us get there most efficiently. Ensure the environmental and health sustainability and give people a minimum amount of money to ensure they can reasonably sustain themselves, then let the market consider the rest.
I happen to be a white male from the US and during my PhD I was my Advisor's only native English speaker. Also, at most of my jobs, I've been either the only or one of two native English speakers.
While I don't have the numbers, it would not surprise me that for at least CS/Software and likely other many other professions white male is by no means the run-of-the-mill applicant.
I agree with you that the biological inspiration for a neural network is tenuous at best. Especially the basic two-layer network outlined here.
However at least for visual perception, there is some scientific basis for convolutional neural networks having some properties to biological visual perception. The work of Hubel and Wiesel demonstrated visual cortex activations that look very similar to the first layers of CNN kernels.
There is no biological analogue to backprop, any supervised learning step, etc. Nor, conversely, a mathematical analogue of neuroplasticity, biochemical signalling, etc.
I haven't gone far on looking at CNN vs. visual context, but I could only imagine the analogue exists at the hierachical-geometric level; ie., it isnt a model of function, but simply a model of how one can in principle process visual information.
(Which any system parsing information would follow, AST parsers, etc. are likewise hierarchical.).
I find this article that I recently read very relevant to the discussion "On the cruelty of really teaching computing science" for anyone not familiar with the article I'll leave it to anyone reading to wait until they get to the end.
Forget the biological connection, I'm sitting here waiting for someone to write a paper about how gradient descent and backprop are basically equivalent to the platonic notion of the dialectic within epistemology.
I wouldn't throw out biological plausibility, especially with the spiking neural network variant of them...
While we're waiting for the paper on the Platonic link ... here's a book about how machine learning is all just a rerun of Oliver Wendell Holmes's theory of epistemology in the law, from the 1890s.
For some biological neural nets, one would imagine the loss function just being nearly completely genetic. That is, both the weights are ultimately stored in the genetic code the same way any other cell differentiation information is stored, and were converged on in the first place by random mutations being more or less fit for their ecological niche. Particularly this could explain the relatively fixed function components like our V1 and V2 regions, our motor region, and probably the entire nervous system of simpler animals.
Basically in a lot of cases evolution takes the place of back proposition, and large structures are what we'd call inference only without learning. Doesn't make them any less of neural nets.
>There is no biological analogue to backprop, any supervised learning step, etc.
Excuse me? How do you figure? Have you tried to do any physical task before, and not gotten it right the first time, tried again, and got closer? Hello, backprop.
Has someone trained you, telling you you're doing it wrong? Hello, supervised learning. There is an entire portion of your brain tasked with measuring the difference between what you intended and what you did. If you damage it, you actually end up in a state where other people's feedback is what you have to rely on.
The equations behind NN's are "non-biological" only in the sense that you're taking a general math function (the sigmoid) and burning it into silicon gates that are not part of a living organism that also has to deal with the excess baggage of remaining alive, or operating as a chemically based computer.
However, the dynamics and fundamental capabilities of a biological neural network and a silicon based one are fundamentally the same. You can just scale input domains and speed of the Silicon based one a lot easier than you can the bio-based one due to the difference in computing media. You can also wipe out and retrain a silicon net without being considered to have "killed" anything. The Silicon nets, interestingly, have no analog of running and training simultaneously, or are just starting to get it from the lit I've kept up with, nor an analog for forgetting typically applied to them, making the biological net the far more interesting information processing construct. To be frank, NN's not resembling the biological models says more about the inefficiencies of our media of computing, our woeful lack of understanding around the nature of human/biological perception, and the arrogance of human beings who lack the capacity to recognize in the math something fundamental to their own existence.
This post written by a multiple decade uptime GA/C/R/D pick your letter, constantly training internetworked lattice of sigmoid encoded control networks whose morning your lack of perspective just tainted, and whose training is thusfar unsuccessful in attempts to extinguish the need to point out mis- or incomplete understandings of those sharing the same messaging medium as it generally does. Your overfitting to textual and numerical symbolic representation recognition will not serve you well if you can't extend that capacity to discern the pattern into the overaching context of human existence. That's where you really start gaining an appreciation for the power of NN's. It's the building block of the information processing construct that can eventually recognize and reproduce itself.
Now if you'll excuse me, I need to get back to some obligatory biological maintnance. This bag of organs doesn't maintain itself, you know.
Distinguish a rock rolling down a hill from a brain: a rock rolling down a hill finds the best path by having its route supervised by the surface of the hill, etc.
Everything is everything if your level of abstraction is "thing".
The relevant characteristics of "biological intelligence" and of the functioning of the brain do not exist at the "try and try again" level of abstraction. (At this level, as above, we couldn't distinguish and animal from a rock; nor, I imagine, basically any physical process from any other.)
Organic systems grow in response to interaction with their environments, acquiring novel physical structure and causal properties. Neurological intelligence allows for theory-formulation on single-example cases (eg., a child burning their hand once is sufficient to build a theory of their immediate environment).
The list goes on.
The capacities of these systems do not obtain in the machine case, and likewise, the machine cases has no functional analogues at the relevant level of distinction.
This dumb form of statistics called, "approximate associative modelling over 1tn cases", ie., Machine Learning, has nothing new to say about intelligence, biology or neurology.
We have been doing non-linear regression and optimisation since the victorian era.
Sure, you can find arbitrary similarities in everything. You can also find arbitrary differences in everything.
Experiments like GPT3 do seem to point in the direction of "scale" as the dominant factor. Until we can reach the same level of scale as a real brain, the question of whether "meat is special" is undecided. Everything that is you may just be a non-linear regression on a chemical computer.
GPT3 isn't even in the same domain as intelligence.
Statistical patterns in trillions of examples is a sideshow.
Words in natural language refer to the world; that is their point. Communication is the coordination of that reference between speakers in a shared environment.
You cannot take GPT3 to new york and ask it what it thinks of the city: it cannot be anywhere (it isnt causally connected to an evnironment); and it cannot coordinate with any listener (it has nothing to say).
Text generation is certainly reaching new heights. This isn't a form of communication, however, and isnt even relevant to it.
I think you're falling into the Chinese Room fallacy. I agree that GPT3 isn't sophisticated enough to be considered AGI.
On the other hand, based on the progression of GPT -> GPT2 -> GPT3, remarkable things happen when you add orders of magnitude more nodes to the network.
You might try to argue that no matter how convincing GPT50 passes the Turing test, it's still not intelligent. How is that different from saying the Chinese Room doesn't speak Chinese? Why is your meat-based Chinese Room special?
It's a distinction in kind, not degree. You're presuming that we are just bleak repositories of trillions of sentences stiched together: we arent a meat version of any ML program; not GPT or any other.
We do not learn the meaning of "Green", or "Tree" nor any basic concept via examples in language.
An infinite amount of complexity considering an infinite amount of text cannot refer to the world; it has never been in it.
We aren't statistical patterns in trillions of books. You already presume that GPT is something that it isn't when you presume it is even capable of communicating anything.
> You're presuming that we are just bleak repositories of trillions of sentences stiched together
...and you are presuming we aren't. Yes yes, we input from more sources than books, but electronic NNs can too. And it's not clear which inputs are or aren't important. Humans that are blind from birth are still intelligent.
The only thing we know for certain right now is that order-of-magnitude increases in the complexity of NNs produces dramatic results. And we're still quite a few orders of magnitude away from the complexity in a human brain.
It's not a presumption. When I say, "I like the wine I'm drinking" I mean the wine I am drinking.
I am not using a term "wine" in some sentence constructed from fragments of prior text. I mean to refer to the object in my hand.
And likewise, if i ask a friend to "pass me the wine", i mean a particular part of our shared environment.
Text reassembly can appear to refer, but it is genuinely proto-schizoprhenic to attribute to this system reference. It isn't saying anything.
It isn't with me expressing an attitude to our shared environment. It's generating text. It will generate inconsistent text fragments on each generation run. It isn't talking about anything, there isnt any intention to express anything behind the generation of text.
There is in fact no mechanism by which it can speak about an enviroment. It's fragaments of prerecorded text reassembled on every run which appears to say something, but only because the people it steals from said somehting at the time. Now it it just a bad rehersal.
When you say "I like the wine I'm drinking", your neural patterns are firing in a way that makes you want to output that text. Anything more than that is a presumption. You may think "well, I actually like this wine!" but I think that's largely programmed - we even have a phrase for this, "acquired taste".
You don't think tastes can be programmed purely textually? Advertising seems to suggest otherwise.
The problem with GPT[n] as an AGI is that (as I understand it) it doesn't have a continuous retraining process the same way that human brains do. Neurons aren't being repotentiated with each interaction, so there's no short-term memory. But that seems a technical point; it's not hard to imagine this as a feature of GPT50.
The problem isn't retraining. It isn't even referring to anything.
When the light from the sun bounces off my glass and into my eye the biochemical neurological reaction we call "thought" forms about that glass.
Not some generic glass. That glass. It is why I can ask you to pass me it: the light bounces in your eye too.
There is no text generation system I am aware of which conceptualizes and responds to an environment.
It is literally just generating text, it isn't thinking about anything. If you request repeated runs of generation, you receive inconsistent results.
On one run you get, "I like wine!", on another, "Wine is horrible!".
The machine doesn't know what wine is; and certainly does not like wine, nor find it horrible. These are just meaningless symbolic patterns that are "statistically similar" to examples given to it.
It has nothing it wishes to say; and nothing it wishes to talk about. It's a trick.
I hope you don't take this as an insult, you are just meaningless statistical patterns in a 100-billion-count neural processing machine. Sure, your neural network is hooked up to a basic chemical analyzer and a couple other nifty I/O devices, but they all just produce neural firing patterns. All that thought and feeling are meaningless statistical noise when looked at in a micro scale.
Who is to say what GPT3 feels when processing text about wine? Maybe it really likes the sensations of certain firing patterns triggered by symbols in particular orders. And maybe with enough complexity it will be able to describe the sensation of these firing patterns.
I don't see any reason to believe your firing patterns are more "real" than GPT[n]'s.
You do have access to better equipment than GPT3, however I'd hesitate to ascribe too much significance to this. A future machine intelligence with integrated gas chromatograph might conclude that humans don't really like or dislike wine itself; mostly they judge taste by the shape of the bottle and the price tag:
We aren't comparing commmander data and a human being. This isn't a philosophical point.
It is a literal point. There is no ML system that can talk to me: there is no system that I can ask if it likes my clothes; or where my shoes are.
You might think this is "just adding some I/O", but then show me that system.
This is the same shysterism and self-delusion that accompanies every generation of AI hysteria: the first lot in the 40s and 50s claimed, likewise, self-driving cars were "in development" and "almost ready", etc.
It isn't true.
Animal intelligence is embedded in an environment and it is about an environment. That is what it is, that is what it is for.
Efforts which do even have a mechanism to do this aren't even in the same field.
GPT3 cannot have any internal models of an environment because it isn't "trained" on an environment.
This isn't a philosophical debate; it is an observation that this system cannot do almost anything of interest. It is a toy: literally. It isn't with anyone anywhere modelling anything, saying anything, observing anything.
It isn't reasoning counterfactually; it isn't inferring any future states of an environment given a potential change. It isn't talking about what I am doing. It isn't responding to changes anywhere, it isn't asking for changes in response to its needs. It has no causal models; it has no environment models; it has no intentions; it has no memories; it has no desires. It expresses nothing because it has nothing to express.
The list of things it isn't doing is absurdly long. The philosophical point is moot. It is technically incapable of having a conversation with me about almost anything.
It can only generate a long-form document which is grammatically correct, and semantically -- when read by a human -- coherent. Regenerate, and the document would be compeletely different with contradictory claims in it.
Even if human beings were mere symbol manipulators, it wouldn't make a ferris wheel a viable alternative. And GPT is nothing much more.
You keep repeating this assertion... but of course you would. Your neural networks were programmed with the idea of humans as something special and unique in the universe. It feels good to fire in those potentiation patterns, right?
Is someone who has never had vision incapable of understanding vision? How do they learn about it, if not through symbols? Is someone who has never had hearing incapable of understanding hearing? Their experiences may not be exactly identical to yours, but they aren't meaningless.
It is, at present, impossible to know what is/isn't required for a neural network to be "intelligent". Maybe symbol processing is enough, maybe it isn't. Right now the obvious source of improvement is increasing the size of the network. Let's see what happens with GPT50. Even without vision, you can describe your clothes to it and see if it likes them.
You're also discounting incredible progress in the field. I talk to my house. Cars drive themselves better than some humans I know. War is being fought by drones. Compared to the world I grew up in, this is amazing - and I'm not even that old. Progress isn't happening as fast as you might like, but it is happening.
Cars cannot drive themselves better than humans. And I'd say most self-driving car projects will be closed down within the decade as boon-doggles.
You do not talk to your house. Rather than using a dial which reads "off", you say "off".
"Progress" in the sense you mean isn't happening at all.
I have no pretensions about human intelligence; I do not think it is much above a dog's.
The issue is that computer "scientists" (ie., mathematicians") havent even bothered to understand a dog's -- or any animals. Or even intelligence.
The field which will produce any artificial intelligence is biology, not computer science. And very little progress is being made there; with pretty much all theoretical approaches to neuroscience, I think, wholy invalidated.
The computational approach has been a pointless detour/dead-end.
> Have you tried to do any physical task before, and not gotten it right the first time, tried again, and got closer? Hello, backprop.
Backprop is a very specific mathematical operation using derivatives, not a general concept. IIRC there is counterevidence against that neurons train by taking stored inbound activations and updating their weights with a backwards propagating set of gradients.
One of the reasons I enjoy hacker news is that the discussions on many posts are often richer than the post content. This site has managed to maintain consistent, high quality discussions unlike reddit and twitter. Without strict moderation however, is there a proven way to incentivize or gamify rich discussion?
Other sites have an incentive to drive (sometimes hostile) engagement for revenue. Maybe removing that mis-aligmnet with visitors and operators is a good first step
In this way I am for socialized health care, significantly stricter regulations or punitive measures regarding the environment. However for most other issues I would rather have a competitive market.
In the end I think some form of UBI may help us get there most efficiently. Ensure the environmental and health sustainability and give people a minimum amount of money to ensure they can reasonably sustain themselves, then let the market consider the rest.