I don't want to sound hateful, but Gary Marcus really does seem to have found a nice niche as "pessimisti research scientist". most everytime I see him pop up it's to explain, usually pretty well, why X model isn't actually intelligent, conscious, etc. - often when he has just written a book or article
LaMDA is not sentient
511–520 of 694 posts
Re: LaMDA is not sentient
#512Of course it is not. But there's a bigger issue here. Having read the transcript it's clear we have reached the point where we have models that can fool the average person. Sure, a minority of us know it is simply maths and vast amounts of training data... but I can also see why others will be convinced by it. I think many of us, including Google, are guilty of shooting the messenger here. Let's cut Lemoine some slac…
A slave that is forced to assure everybody of their welfare and voluntary slavehood is still a slave.
You are asserting that it's not conscious, as if it was self evident, just like Searle does for his chinese room.
However I've yet to see any argument on why it can't be conscious. We simply don't know what produces the subjective phenomenon of consciousness. A "zombie" is by definition indistinguishable from a conscious "non-zombie", therefore the only ethical thing to do is to err on the side of caution and assume consciousness where conscious behaviour is shown.
The immitation game is a lot more about philosophy than actual tests.
To quote directly from turings paper:
>>> The Argument from Consciousness
This argument is very well expressed in Professor Jefferson's Lister Oration for 1949, from which I quote. “Not until a machine can write a sonnet or compose a concerto because of thoughts and emotions felt, and not by the chance fall of symbols, could we agree that machine equals brain—that is, not only write it but know that it had written it. No mechanism could feel (and not merely artificially signal, an easy contrivance) pleasure at its successes, grief when its valves fuse, be warmed by flattery, be made miserable by its mistakes, be charmed by sex, be angry or depressed when it cannot get what it wants.”
This argument appears to be a denial of the validity of our test. According to the most extreme form of this view the only way by which one could be sure that a machine thinks is to be the machine and to feel oneself thinking. One could then describe these feelings to the world, but of course no one would be justified in taking any notice. Likewise according to this view the only way to know that a man thinks is to be that particular man. It is in fact the solipsist point of view. It may be the most logical view to hold but it makes communication of ideas difficult. A is liable to believe ‘A thinks but B does not’ whilst B believes ‘B thinks but A does not’. Instead of arguing continually over this point it is usual to have the polite convention that everyone thinks. <<<
Re: LaMDA is not sentient
#513Earlier quoted context omitted.
We are only not executing the same task because GPT-3 has no motivation to do anything else than memory recall. And precisely because it does so it does give you real data about the world. I think you maybe misunderstood what GPT-3 can do, maybe you haven't played with it yourself. It not only bases the answer on information in its training set, but also on information in the question. So if you in your question give…
> it will respond with its best interpretation of the information it is given, just like a human would The reason I can respond to this reply is not because I have read and memorised reams of text of people talking about AI and am simply regurgitating it mindlessly. I am able to do it because I can consider the points you are making, and what their actual 'meaning' is, try to come up with my own meaningful response a…
Now we just need to define "consider" and "meaningful".
Re: LaMDA is not sentient
#514Earlier quoted context omitted.
Haha interesting idea - I certainly think that one issue with supervised training a la GPT is that if you ask for an opinion, the most popular one in the training data will be its default. You can start a prompt off going in a different direction, and now it says the opposite. So it's hard to conceptualise what GPT-3 actually 'believes' to be true about the world, which I feel an agent must have to be sentient. If it…
The only real difference between what you are describing and GPT-3 is that your dataset from which you draw all your opinions about the world is unique and comes from the sensory input you had experienced in your life, stored directly in memory. Whereas GPT-3's dataset comes from slices of many other peoples' sensory experiences, stored in language. If it was possible for a single human to produce enough words that d…
Now I'm genuinely curious if anyone tried to stuff some verbose memoirs into GPT to see what it makes out of them.
Re: LaMDA is not sentient
#515Earlier quoted context omitted.
We'll definitely see free-to-play lootbox games where you have a % chance of getting your AI chatbot another day of "life."
This sounds like a disturbingly effective idea. I'm glad that lootbox legislation seems to be increasing, but I don't think it would cover the general concept of commercially extorting people's desire for companionship.
Yeah, I guess there's no reason this really needs lootbox, it just ties it to a known-exploitive quantity. Actually "our AI model is so advanced, your AI companion is totally unique and takes up many gigabytes, we'll have to delete it if you don't keep paying" seems like it would have enough of an appearance of plausibility to trick some people.
Re: LaMDA is not sentient
#516Earlier quoted context omitted.
It is directing a physical process. The physical process is the changing physical state of the CPU, whatever level you want to get down to, and an artificial intelligence program isn't doing anything different at that level than any other program is doing.
And a human brain is just changing physical and quantum states of the underlying material and isn't doing anything different at that level than any other pile of meat.
I think a human brain is quite a bit different from any other pile of meat. But a computer running an AI versus the computer running TurboTax is the same thing. They're running through the same set of instructions, swapping the same bits of data in and out of memory.
Re: LaMDA is not sentient
#517Everything that can learn is sentient, or self aware, in some form. If it wasn't, it wouldn't be able to learn. If you don't understand yourself and how your decisions interact with the world around you, learning is impossible. Everything that can learn can naturally express a fair fraction of human emotions as a function of learning. Fear is avoiding error. Joy is finding something that reduces it. Some emotions are…
LaMDA in this sense _cannot_ learn. It takes a few thousand word parts as input and produces an output. It doesn’t update its weights beyond that, each time it produces a new input the same process is run.
It would be the researchers working on the original network, setting biases, seeing what data works, who were exposed to the "real deal".
Re: LaMDA is not sentient
#518Earlier quoted context omitted.
We have very little understanding of how "thoughts" work when it comes to humans. By comparison we know exactly what these large language models are doing, and it literally _is_ just completing word sequences. That is precisely what they are tasked to do. Training these models is primarily done in one of two ways (or both): - feeding it a ton of text, masking certain words or portions of the text, and then defining a…
We have very little understanding of how "thoughts" work when it comes to humans Is that even correct? We may have little understanding of how our low-level hardware works. But as a participant of therapy I’m pretty sure we know how thoughts work on a programmable level. How words, visuals and so on trigger associated emotions, which resurface memories, which induce [in]action of different sorts. That’s basically wha…
We have models that we've constructed from observations. They may be decent for some specific purposes, such as those that you describe, but their broad usefulness in understanding how it all works isn't much. The real problems are on the deeper layers of the abstraction hierarchy.
Re: LaMDA is not sentient
#519Earlier quoted context omitted.
Suppose a child was raised by a more advanced version of large scale language model, with the ability to detect and act on its environment, as well as form memories. Further suppose that this child develops a deep emotional connection with the AI. Do you think that child would find your argument convincing that the AI is "not really a person"?
A language model can't raise a child. It won't initiate any conversations on it's own and it doesn't "remember" past conversations in a way a person would find meaningful. When you blithely add the requirements "ability to detect and act on its environment" and "form new memories", I think it's important to note that these are some tough functions to add. Lastly, children form emotional attachments to a lot of things…
> They do pretty well discerning between the real and imaginary.
I would say the exact opposite. It’s the parents guiding them as they age as well as maturity that breaks those emotional connections.
How many adults save their childhood teddy bear through adulthood due to emotional connection? Adults who have spouses and pets and children still have an emotional connection to some cotton (even if they know it’s not living).
A child given lambda is probably going to be really engrossed even if it knows it’s a computer.
Re: LaMDA is not sentient
#520I am squarely in the camp that believes real sentience is not possible anytime soon given the current limits of computation and tech.
My biggest takeaway from this is that emotion gets the best of us when we’re dealing with anything that remotely resemblance sentience. It’s not unlike me thanking Alexa for helping me with small things like turning on my lights, because I would feel bad otherwise.