▲ 108 points
back
108 comments
1879
These arguments aren't new.
This is good argument for why tech/science/stem need to read history, or 'older' science history. To at least be familiar with the science that has come before, either to know what was discarded, to be familiar with past dead ends, or forgotten knowledge that could still have value.
A lot of the AI spurred arguments in the last year seem like so much re-hashing since they were being discussed in 1879.
Absolutely agreed. I find it embarrassing how STEM devalued non-STEM studies for years, and now that LLMs are center stage, so is the near-universal unpreparedness of interested parties and stakeholders to wrestle with issues of cognition, consciousness, and ethics. This, despite literally thousands of years of prior studies on these matters.
Even now, I see basic refusal to engage with this extant literature, and a strong preference to just making shit up as if it's the first time anyone has anything interesting to say on these matters. Again, embarrassing. The industry is totally unprepared on an intellectual basis to wrestle with these issues.
This made me go look up whether STEM includes psychology and medicine. Wikipedia says "depends how you ask." In any case, William James trained to be a doctor and is considered the father of American psychology. If you've read any of his stuff, he was certainly not science illiterate.
Which is not to say I disagree with your point. However, it turns out a lot of people, especially the really smart ones we've all heard about, aren't easily characterized as STEM or non-STEM. That includes all the AI researchers I know.
This is because the humanities got taken over in some sense and filled with mumbo jumbo. Technical people don't want to touch that with a ten foot pole.
Besides, science used to be very much a "thinkers" only thing, because only those people went into studies instead of just being laborers. These were people that would question and ponder the things you mention. These days, STEM is filled with factory widget repeaters that are taught to regurgitate the syllabus and apply it in context so they can be good workers.
> STEM devalued non-STEM studies for years
What does this refer to?
LLMs should requalify Liberal Arts education. The people who can ask questions and understand answers are quite relevant now, seeing how STEM worldview is in many ways no longer relevant.
SDEs should double down on soft skills now, since these are as useful with LLMs as with people.
Are the STEM educated thinkers publishing AI doomsday think pieces or is it the supposedly enlightened non-STEM journalists?
> These arguments aren't new. This is good argument for why tech/science/stem need to read history, or 'older' science history.
Sure, but it is also important not to lend too much credence to these arguments, because they were devised from a position of ignorance. In particular, in 1879 we didn't have neurobiology, nor the theory of computation and complexity, and we barely had evolution. OoS had been published just 20 years before, and molecular biology was many decades in the future. There is value in studying Ptolemy and Aristotle too, but mainly to understand how they got things wrong.
I agree.
You can't take something from 1800's as gospel, like that was the final word and we shall not progress.
Just saying, shouldn't ignore it either.
It is ok to study Plato or Leibniz, and realize it is historical background, and might spur new thoughts. We don't worship Plato as if that was the end of all thought. But lets not forget it, ignore it like it didn't happen, like all ideas today are new.
Agree with you except that the value in studying ancient thinkers and intellectuals is also in recognising patterns and seeing how they probably recognised patterns too.
> forgotten knowledge that could still have value
It's always fun to mention that many of the foundational ideas in the modern wave of machine learning & neural networks stem from work done in the 50s and 60s: perceptrons, backpropagation, stochastic approximation, etc. were all explored in depth back then.
It was only after compute power scaled up enough to apply these techniques practically that they became revolutionary. Really makes you wonder what things people are working on right now that will also need to wait 30 years.
I recently stumbled across a speech from 1855 by an American physician [1], and I suppose proto-electrical-engineer, talking about the telegraph. In it, he makes a direct analogy between the telegraph and the animal nervous system. (People were aware by then, that the nervous system was partly electrical in nature.)
The complexity of the nervous system and its ability to coordinate across distance, allows it to support more complex and bigger organisms, whether that organism is biological or political. Continuing with his analogy, the railroads and canals are the arteries and veins, supplying the raw materials and finished products, etc.
More tenuously, he also appears to make a kind of proto-evolutionary analogy (Darwin hadn't published but Lamarck had published decades before, by this point). And I even detect small hints of an antecedent to the concept of emergent complexity, and the notion of shared structures across biology, sociology, etc.
> The electric telegraph is thus the nervous system of this nation and of modern society by no figure of speech, by no distant analogy. Its wires spread like nerves over the surface of the land, interlinking distant parts, and making possible a perpetually higher co-operation among men, and higher social forms than have hitherto existed. [...]
> We are thus conducted to the result of the highest philosophy: that society in its form of organization, is human, and that it presents in its progressive development continually higher analogies with the laws of individual being. [...]
[1] https://collections.nlm.nih.gov/catalog/nlm:nlmuid-101620765...
I wonder if Channing would be gratified to see the telegraph analogy survived to find new expression in 1970s popular culture:
https://www.youtube.com/watch?v=ivk_irrH1WY
which means that there's probably a bunch of Gen Xers with this song still in their head somewhere (and since it's pretty catchy, some of HNs audience now).
Very true! I think it’s eye opening that almost every one of Chomsky’s lectures on such topics (I.e. “what kind of creatures are we?”) start with a discussion of what Newton, Galileo, or Descartes has to say about it.
The great thinkers of the past aren’t necessarily outdated, in some ways they might still have things to teach us from beyond the grave that we didn’t really appreciate when they were alive.
> These arguments aren't new.
True; the 1768-1774 https://en.wikipedia.org/wiki/Jaquet-Droz_automata were implemented specifically because writing, music, and art were at the time accomplishments that distinguished upper classes from lower.
(their biggest weakness is that being analogue instead of digital, their programming is very sensitive to environmental changes)
Yes, we are and some observable Automatisms are hyper creepy. And if you become aware of them it turns society into a collection of walking grandfather clocks in some aspects. Not a recommendable direction to move, madness is preferable.
An old Fritz Leiber novel, You're All Alone, aka The Sinful Ones, is about a man who discovers that most of the people in the world are automata, and he has been functioning as one for some time.
When people say we’re nothing but computers, neurons firing, chemical reactions, etc they think they’re saying that humans aren’t special.
But I take it to mean that the rest of the world is way more special than we’ve been giving it credit for.
By this point I think the "nothing but" part is the bit that's actually wrong.
It's about putting yourself up on a pedestal while downplaying others.
Of course, at some point those "others" (be it peoples, genders, creatures, or machines) turn out to be as good or better at certain things than you are. And what do you do then? You won't have anywhere to go psychologically without devaluing yourself.
So then you end up shutting down avenues of thought that might otherwise have been productive.
Yes exactly. It’s hierarchical thinking from an anthropocentric pov. IMO the longer I live and the more I learn, the more I see hierarchical thinking as the biggest obstacle towards learning and understanding.
At that point it’s just semantics and relative comparison, I think.
Sometimes I entertain the idea that inanimate objects have consciousness too and it's pretty disturbing because it would mean we are doing pretty much everything wrong.
Well I wouldn't worry much about inanimate objects (especially looking at how we treat animals already).
But then consider higher order systems like a forest or human society itself. Can those have consciousness or at least some sense of experience?
The keyword to look up is "panpsychism" - it was discussed a few times in the past year:
>Is Consciousness Part of the Fabric of the Universe?
https://www.scientificamerican.com/article/is-consciousness-...
>Scientific American on a philosophical grift: panpsychism
https://whyevolutionistrue.com/2023/10/01/scientific-america...
>Is the Universe conscious? A panpsychism Q&A with philosopher Philip Goff
https://bigthink.com/13-8/panpsychism-universe-purpose-philo...
You mean we’re killing things?
Do we even have consciousness or merely the illusion of it?
I would strongly recommend the reading of Huxley's address^1 before reading James's response. The difference is palpable.
---
It was kind of funny to me that Dostoevsky coincided with biology development which made him ponder painfully if we are just reactions of our neuron's "tails" and that there is nothing underlying (such as Soul and God).
Did not expect to see William James pop up on HN.
If you're interested in philosophy (especially epistemology), the pragmatists were a fascinating offshoot.
https://en.m.wikipedia.org/wiki/William_James
IMHO they were a reasonable response to the irreducible complexity that accreted as a result of applying increasingly specific and complex empiricism to questions far broader than it might be suited to. (E.g. How do we know anything?)
Both James and Pragmatic Philosophy have had their own program on the BBC's In Our Time radio show. Very good.
https://www.braggoscope.com/2010/05/13/william-jamess-the-va...
The Metaphysical Club is an excellent book about how the pragmatists shaped all of American thought, too.
Obviously doesn’t replace reading the source material but it does articulate how ideas become encoded into a society.
It's been a minute now, but I'm pretty sure that was on the reading list for my undergrad pragmatism course. Enjoyed it as well!
PS: Hello to another Ethan, although we seem much more common these days.
The Varieties Of Religious Experience was a great read for me
Slightly tangential, but is anyone aware of an app/site that can take a URL like this one and convert the text to a format and layout that's easier on the eyes and easier to read in general? Like Kindle-fy a web page full of text?
Most browsers have a reader mode. Firefox has one built in, I just click the icon to at the right end of the URL bar and it does it nicely for me.
Try the reader view in chrome or Firefox. For chrome on android I automatically got a prompt to view simplified page
Wow, this works great! On Android I recall getting the "simplified page" prompt but never found that to be all that great of a solution but using reader mode with Chrome is pretty neat. You can format it to basically be the same as reading a book from the Play store.
One nit is that if I tab away from the page and tab back the highlighted area is wiped from the reader pane -- assuming I had unhighlighted it in order to start reading.
In Chrome it seems like clicking the pop up is the only way to access simplified view. If you exit, follow a link, tab away, whatever... I can't figure out how to get it back! Any tips? Really confusing that I'm either missing something obvious or the explicit option is missing.
On iOS you can click the icon to the left of the URL input and select reader viewer as well. Hides ads too.
Safari, Firefox, and Edge all have a built-in Reader mode that doesn't exactly the thing you want for the device that you are on. Lots of extensions do this as well. There's used to be a handful of websites but it seems they no longer exist (they were also commonly used to bypass paywalls).
You can use browser extensions to send articles (blog posts etc.) directly to kindle.
I am on mobile app at the moment so I don't have links for those extensions.
I tried to read this but I can’t deal with the writing quality. I think it’s partly the historical nature of the piece, but James is a terrible writer. He is obtuse and pompous and clouds his meaning with confusing sentences. Even the first sentence is a nightmare to read. I’ll try again when I have more focus!
And yes, we have no free will. Not sure if that’s related to this piece.
You just have bad taste.
It’s terrible writing AND I have bad taste*
Fwiw nothing with outside influence is automata as the true randomness from quantum phenomena propagates.
Eg. If you take in real world visual input to a system that from that point on is deterministic the output of that system is now non-deterministic. Similar to how Shrodinger talked about the macro scale cat being alive and dead at the same time.
So the arguments that the neurons in the brain are deterministic or that the transistors in a computer are deterministic breaks down at the first introduction of real world input. The outputs of those systems actually aren't deterministic at all.
As far as we've been able to determine, almost all macroscopic objects can be behaviorally described in terms of classical physics, and this includes neurons.
Even in rare cases where quantum effects are seen, such as photosynthesis, or radioactive decay, it's extremely localized. The fact that a plant's efficiency of conversion of light to chemical energy appears to use quantum effects doesn't affect the rest of the plant or turn it into some "Schrodringer's cabbage" thought experiment. We can still understand the rest of the plant in classical terms.
Even if it were to be discovered that neurons are sometimes (or even always for that matter) behaving randomly due to quantum effects, that would not mean that we have free will and are not automata. It would just mean we are non-deterministic automata.
As said before in this thread, that won't give you free will, and elsewhere on HN, that you won't even get an uncaused cause. Think of your brain as a cryptographic hash function running on a non-reversible computer mixing the quantum noise into its state if you want, so you get your avalanche effect and chaos theory stuff (though I think that's majorly overstating it, people don't behave that erratically). See Good and Real by Gary L. Drescher (2006), but also see my comment at https://news.ycombinator.com/item?id=38789146
I see no reason why that can't be modeled as an automata? Nondeterministic automata are still automata. I think your meanings of 'deterministic' are confused- it's about impulse response. A little bit of stochasticity in the thermal voltage doesn't ruin the determinacy of a computers ALU operational modes.
Deterministic systems, or at least highly predictable ones, can arise for non-deterministic components, as for example the Boltzmann equation demonstrates. Other examples include the models that we build of automata on machines made of matter made of non-deterministic particles.