Hacker News Reader

Comment by swyx | Hacker News Reader

this is a wonderful race: as the machines become more human, the humans are forced to introspect and double down on what it means to be human. One could almost call it the human race. (ba dum tss)

I've had my own problems with proving my own humanity[0]. With this AI wave, I also took a stab at enumerating what machines/AI can't do: https://lspace.swyx.io/p/agi-hard

- Empathy and Theory of Mind (building up an accurate mental model of conversational participants, what they know and how they react)

- Understand the meaning of art rather than the content of art

- Physics and Conceptual intuition

Another related paper readers might like is Melanie Mitchell's Why AI is Harder than We Think: https://arxiv.org/pdf/2104.12871.pdf

(sidenote, always love Maggie's illustrations, it is a real super in a world of text and ai art)

0: https://www.swyx.io/proving-our-humanity

View on HN

As I say in this blog post [1], we are the ones becoming more computerized and limiting ourselves of expressing our humanity through digital means.

There is no race to become more human, just close the computer and go outside and an AI will never be able to compete with you in that regard.

[1] : https://but-her-flies.bearblog.dev/humans-arent-text/

I really like this post. Few months ago I've read a book that really made me finally get a word for what I think we both see happening. it's from a book called Seeing Like a State, and the word is over-abstraction. If you want to know what this is. Imagine a forest with all of it's unpredictable branches, grazing animals, various species of leaves growing from the ground. Now over-abstraction is when you make a plot of monoculture on which not even insects roam and conclude that's plantlife. And I kinda realize that's something that we humans do to other people too. We've mind ourselves kinda predictable and boring because how we express and think by ourselves is in these boxed in, unecessarily fixed, ways. I feel AI is kinda an outgrouth of that too. Since in many ways what makes software AI instead of regular software is how little control you have over it. Not really what it does.

AI worries me though, not because I believe it will be intelligent or sentient or whatever anytime soon. But because it cuts people who do important work from money. Which means there's a high chance we'll be poorer off if we don't do something about it in the near future.

> go outside and an AI will never be able to compete with you in that regard.

Why not?

> humans aren't text

Was Helen Keller human?

Re. your list: plenty of people have problems with those.

I frequently point out that $ThingWeSayIsntAi isn't as good at something as a person who is good at it, but it is rapidly becoming better than a person who isn't good at it. Coming from decades of systems pretending to be remotely competent, this is a striking inflection point. The Times recently cooked a Thanksgiving dinner based on ChatGPT recipes. It wasn't very good, and they closed by saying "I guess we still have jobs!" People don't grock exponential growth.

The fact that we are still litigating whether authorities responded too lax, adequately or too harsh three years after the start of the COVID pandemic is proof positive that you are 100% right.

It's interesting how fast this development is going but I fear that with all of the other stuff going on in the world and the fact that we have barely managed to get a grip on what it means to have a free of charge pipe between a very large fraction of the world population, that we are in for a very rough ride. The various SF writers that addressed the singularity were on the money with respect to our inability to adapt, they were too pessimistic about the timetable. The ramp-up is here, whether we like it or not and the only means that we have at our disposal to limit the impact a bit is the rulebook. But then it's a huge game of prisoners dilemma, the first one to defect stands a good chance of winning the pot.

One more thing that can help: the same tool that gives can take away: AI can help to figure out which art/text/music was generated by AI and which by a human. Someone else in another thread earlier on HN made the comparison between pre-AI and post-AI art that it is like Low Background Steel (I can dig up the reference if you want), and I think that's really on the money, everything that we made prior to the emergence of generative AI is going to be valued much more than anything that came after unless it is accompanied by a 'making of' video.

https://news.ycombinator.com/item?id=32561868

It is the shitest race ever. Utterly pointless.

Sounds like depression talking. Not everything is shit, but if everything smells like shit, check your shoes.

I mean that in a charitable way. Small depressions can easily become large ones if they are allowed to run amok with your feelings.

HN person diagnosing depression. Well done.

I’m far away from depression. Just think the AI race is absolutely a pile of shit for humanity.

I wish you weren't another one of those people who respond to concern with hostility. It's so boring.

Genuine concern doesn't respond to hostility by calling it "boring", even if it does stoop to the old "check your shoe" canard. Anyway, your whole basis for "concern" is that the person expressed a strong negative opinion. As evidence for depression, that's, ah, weak.

I am expressing my opinion and concern for this pointless race.

As an observer of your exchange of words, I would appreciate more reason behind the opinion because I would like to understand why you feel this way.

We are trying to invent things that mimic humans for the purpose of increasing what? While the world falls into pre climatic disaster.

It’s all very interesting but what utility does it have for humanity?

Do humans understand the meaning of art? Of modern art? Regular humans? Other than the author?

not sure about the meaning, which would imply there'd by an answer to what meaning a certain art peace would give. But a human being will have their own reaction to a peace of art.

For example I've often heard said that great art is something which makes you feel something. A machine cannot feel

Do we actually know the other human feels that something? We don't, because we only hear them pretending they feel and usually believe them. Well, a machine could pretend just the same - they have enough training data to know what feeling is appropriate to claim.

Once I realised I had aphantasia (I don't see things in my "minds eye"), in my 40's, after having my whole life assumed people who said things like "visualise X" meant it abstractly or metaphorically rather than literally, it really drove home how little most of us understand about the inner mental processes of other people.

Even more so seeing people express total disbelief when I explain my aphantasia, or when others point out they don't have an inner monologue or dialogue.

Most people have far less understanding of other peoples inner life than they believe they do (and I have come to assume that applies to myself too - being aware that the experience is more different than I thought just barely scratches the surface of understanding it).

Yes.

Ultimately this is a question of meaning. Where is meaning to be found?

It's going to come as a surprise to many that it is only to be found in the individual. Not in countries, nor in religious groups, or in football teams, or political parties, or any form of collective endeavour. The meaning is inside.

We can't know how other human beings feel, nor can we know whether machines can feel. However, it is a safe bet (to me) that other humans are like me (more or less). And that machinery is inanimate, regardless of appearances.

But then you will get attempts to anthropomorphise machines, eg giving AI citizenship (as per Sofia in the UAE). What is missed with this sort of anthropomorphising is what is actually occurring: the denigration of what it is to be human and to have meaning. A simulacrum is, by definition, not the thing-in-itself, but for nefarious reasons, this line will be heavily blurred. Imo.

We are already drawing lots of lines when antromorphising animals. Does an orangutan give meaning to its drawings? An elephant when it paints? A parakeet or a magpie when decorating their nests? Even fishes do decorations to attract mates, so their mates definitely draw some meaning from those actions. Now if you define "meaning" as something only humans can draw then okay machines won't have that meaning - although we both agreed each human will draw a different meaning anyway. This of course also excludes any sentient aliens from drawing meanings from human art, because well they are not humans. And that we humans will never understand a fish's art because we are not fishes. So meaning is both individual, and species-related? Or either? Which one is now the real meaning, the one the individual draws (then species is not relevant, so why not including machines) or the one the species draws (then it's also a group meaning, so again why not including machines)?

Or maybe your corner stone argument is "machinery is inanimate" - which would be another discussion by itself...

I don't think anthropomorphising animals is in the same category as anthropomorphising inanimate objects. A child might believe their teddy bear to have a character and life, but this is being projected on to the toy by the child. An animal however has its own experience, life, etc. What I've said can be objectively determined, do you agree?

I would agree that animals do have a life, but they are not at the same intellectual level as humans. You mention art though - this is a bad example for me - one that is not clear in meaning to humans. I have my own interpretation of what art is.

But just that - that I have an interpretation of what art is, this is a difference between humans and complex animals. It is evident that we handle complex concepts, and play with them. This is not the case for animals, and if there is some nascent behaviour like this, it is nothing like at the level that humans do.

That covers my views (more or less) on the differences between humans, animals, and inanimate objects (computers, toys).

The real point I was making though, is that meaning resides inside oneself. That is where the experience is 'enlivened'. You can watch cartoons move on a screen, actors move on a screen, other people in real life - but all that is just visual/auditory inputs. What gives it meaning is that you 'observe' this.

I know people talk about AI becoming sentient etc, but to me this is an impossibility. AI can no more become sentient than can the cartoons on the screen, or stones on the beach. AI can however, give the impression of sentience, better than a toy or something like that. But this is not conscious awareness any more than an actor turning to the screen and talking to the viewer is an example of the TV being sentient.

I understand that many scientific people have been trained to objectify themselves, and consider their 'anecdotal experience' as irrelevant or as a rounding error. I think this is a massive error personally, but those with that scientific mindset will not like what I'm saying. There is something special about each individual - the experience of consciousness is infinitely valuable - and although it is possible to conceive of objects doing a passable or great impression of a conscious experience, the difference is akin to seeing a fire on a screen, and experiencing it in person - ie a world of difference.

The discussion was specifically about art, that's why I mentioned art. To come again to my point, a human thinks it's sentient because a human thinks it's sentient (not kidding). We agree that towards the exterior, we can get an illusion of sentience from a TV set. But towards the interior? I only claim my neighbor is sentient because I claim I am sentient and the neighbor is human thus will be sentient as well. I don't have any more access to their sentience than I have access inside the "black box" TV set's sentience. So it all revolves around my own sentience, used as yardstick for all humans and to some extent, animals (plus the old debates about slaves, women, aliens...). I personally think we are all sentient because I think I am sentient. So... if a machine thinks it's sentient, will it be sentient? In a different way? Is there only one sentience? My consciousness is infinitely valuable (to me!) thus any human's will be (maybe less than mine, eh), and a machine's not much (but how much?). Or a rat's? Oh well, biology is one thing, and philosophy is another thing and they're definitely not mapping 1:1.

This is what I said in the beginning, so I think we broadly agree.

> It's going to come as a surprise to many that it is only to be found in the individual. Not in countries, nor in religious groups, or in football teams, or political parties, or any form of collective endeavour. The meaning is inside.

Where you say:

> So... if a machine thinks it's sentient, will it be sentient?

I dispute the assumption you are making. A machine can't think. There is no sentience occurring, even if it appears on the outside that there is.

And also, while it is fair to assume that other creatures that look like you are similar to you, it is not a fair assumption to think all things, inanimate things, have the same degree of sentience as you experience.

My misunderstanding is probably on the very definition of "sentient", which the dictionaries limit to having senses and feelings. Actually not only dictionaries, because according to different legislations quite a few animals are classified as sentient beings. Now a machine will necessarily have senses, while feelings are at the least signaling mechanisms. None of which are really something too fancy, so you must be meaning something else/more to the sentience, something not covered by dictionaries. Something more that "the ineffable" or "je ne sais quoi"...

PS: a machine can't think? Why did you say that? Because it cannot be sentient? Is thinking the difference making one sentient? Then maybe we should analyze "thinking" instead, as it's a term definitely easier to check.

I've no objection to animals being considered sentient, they are independent, I think.

A machine doesn't have senses - it's not alive. It has inputs. I think you are somewhat living within the metaphor that 'we are like computers'. It's a fine metaphor, but you are not in fact a computer (if you are in fact a human, and I cannot tell).

As I have also said, your experience is your own only - you have no way to confirm that other people's experiences are coherent with yours. This is because you only have access to your subjective experience. When looking at the objective world we all experience, it is a fair assumption to believe that the other creatures that are like us also have their own internal experiences, meanings, etc.

I think you would agree that a toy is not sentient. Nor a player piano. No machinery or non living things are sentient, or ever could be. So why would you allow yourself to believe that some machinery is sentient, just because it does a good impression of something that is sentient?

This confusion actually relates so a psychological quirk and misunderstanding we commonly make (and have been taught from birth). We often talk in terms of 'we', eg 'we, the people' - as if one can ever talk for others. We contextualise ourselves as a third party in the objective world, as if we were not having a subjective experience. It's a common use of language (I do it myself) but it is an error, or rather, it is fine for the objective world but does not cover the full range of the human experience. From a subjective perspective, one can only ever talk for oneself, feel for oneself, grasp meaning for oneself. It is illusory to think that anyone else could ever be able to represent you, or that you are able to represent others.

Ultimately, I know this is not an argument I can prove objectively - I'm sure that it is even possible to code an ai to argue as if it did have a rich internal life (which is impossible). Anyone witnessing its argument could agree. But what of it? What if everyone agreed? As I said, meaning resides within the individual, but I think a lot of harm will occur to those with a collective mindset, who will be unable to counter the combative ai in this example - they will cede power, rights, etc to machinery - they will debase themselves.

>It's going to come as a surprise to many that it is only to be found in the individual.

I'm going to give this a YnEoS as an answer.

So, yes, meaning is individual and occurs in your mind.

Also, no, your ideas of what meaning should even be in the first time are affected by your collective endeavor, your political party, your football team, your religious group, and your country. There will be a statistically high correlation with of your views of what meaning is and your affiliations with any of the above parties.

>hat is missed with this sort of anthropomorphising is what is actually occurring: the denigration of what it is to be human and to have meaning. A simulacrum is, by definition, not the thing-in-itself, but for nefarious reasons, this line will be heavily blurred. Imo.

I mean, isn't the meaning of being human to live on the Serengeti plains, fighting for survival and enough food to eat, and everything since then is just the simulacrum? Humans create society which create the simulacrum in the first place. That line was blurred so long ago we have no idea where it even existed.

If anything I would expect machines to be be better at determining people's feelings. Unless you know the person well, you are using things like facial expressions, body language, and tone of voice to figure out how someone is feeling, and hoping that they react in conventional ways.

Now that we've willingly told companies everything about ourselves, for younger people straight from birth sometimes, their machines will be able to use all this context to construct a more accurate picture of how a person might feel about an arbitrary subject.

Everyone knows that famous story about a woman being recommended pregnancy-related products before she even knew she was pregnant herself, and that was before this latest round of AI.

Also: The feelings humans have are also influenced by their culture. Feelings are not only felt but also enacted. And the enactment influences the feeling.

The final scene of midsommar is a great illustration of this.

Post-modern consensus is that there is no meaning.

That sounds like nihilism with extra steps.

I agree. I'm not a nihilist but that's the logical conclusion of the dominant philosophical position.

Appreciation of the complexity of the natural world.

Reply on news.ycombinator.com