site banner

Culture War Roundup for the week of February 20, 2023

This weekly roundup thread is intended for all culture war posts. 'Culture war' is vaguely defined, but it basically means controversial issues that fall along set tribal lines. Arguments over culture war issues generate a lot of heat and little light, and few deeply entrenched people ever change their minds. This thread is for voicing opinions and analyzing the state of the discussion while trying to optimize for light over heat.

Optimistically, we think that engaging with people you disagree with is worth your time, and so is being nice! Pessimistically, there are many dynamics that can lead discussions on Culture War topics to become unproductive. There's a human tendency to divide along tribal lines, praising your ingroup and vilifying your outgroup - and if you think you find it easy to criticize your ingroup, then it may be that your outgroup is not who you think it is. Extremists with opposing positions can feed off each other, highlighting each other's worst points to justify their own angry rhetoric, which becomes in turn a new example of bad behavior for the other side to highlight.

We would like to avoid these negative dynamics. Accordingly, we ask that you do not use this thread for waging the Culture War. Examples of waging the Culture War:

  • Shaming.

  • Attempting to 'build consensus' or enforce ideological conformity.

  • Making sweeping generalizations to vilify a group you dislike.

  • Recruiting for a cause.

  • Posting links that could be summarized as 'Boo outgroup!' Basically, if your content is 'Can you believe what Those People did this week?' then you should either refrain from posting, or do some very patient work to contextualize and/or steel-man the relevant viewpoint.

In general, you should argue to understand, not to win. This thread is not territory to be claimed by one group or another; indeed, the aim is to have many different viewpoints represented here. Thus, we also ask that you follow some guidelines:

  • Speak plainly. Avoid sarcasm and mockery. When disagreeing with someone, state your objections explicitly.

  • Be as precise and charitable as you can. Don't paraphrase unflatteringly.

  • Don't imply that someone said something they did not say, even if you think it follows from what they said.

  • Write like everyone is reading and you want them to be included in the discussion.

On an ad hoc basis, the mods will try to compile a list of the best posts/comments from the previous week, posted in Quality Contribution threads and archived at /r/TheThread. You may nominate a comment for this list by clicking on 'report' at the bottom of the post and typing 'Actually a quality contribution' as the report reason.

15
Jump in the discussion.

No email address required.

Language models aren't sentience. When you ask the AI if it feels pain, and it generates some thoughtful paragraph about it being a machine or whatever, that's not 'the AI' sharing it's thoughts. It's a text completion algorithm generating some text based off human literature about fantasy AI personas waxing philosophical.

And yet it understands.

Someone, I think Bertrand Russell, said we compare the mind to whatever is the most complex machine we know. Clocks, steam engines, telephone relays, digital computers. For AI, it’s the opposite: as capabilities increase, and our understanding of AI systems decreases, the analogies become more and more dismissive.

The mainstream, respectable view is this is not “real understanding”—a goal post currently moving at 0.8c—because understanding requires frames or symbols or logic or some other sad abstraction completely absent from real brains. But what physically realizable Chinese room can do this?

[...] What is left of rationally defensible skepticism? For once I’d like to hear an argument that doesn’t rely on Cartesian dualism, stoner metaphysics, or from people still clinging to GOFAI nostalgia like the Japanese holdouts.

If you’re going to tell me intelligence requires symbolic rules, fine: show me the symbolic version of ChatGPT. If it is truly so unimpressive, then it must be trivial to replicate.

There is a species of denialist for whom no evidence whatever will convince them that a computer is doing anything other than shuffling symbols without understanding them, because “Concepts” and “Ideas” are exclusive to humans (they live in the Leibniz organ, presumably, where they pupate from the black bile). This is incentivized: there is infinite demand for deeply credentialed experts who will tell you that everything is fine, that machines can’t think, that humans are and always will be at the apex, people so commited to human chauvinism they will soon start denying their own sentience because their brains are made of flesh and not Chomsky production rules.

All that’s left of the denialist view is pride and vanity. And vanity will bury us.

I appreciate the response, but again, this isn't sentience. GPT is just text completion. It's not an intelligence, it's a tokenized list of words. Finding links between ideas by consuming the literary cannon of the human race is a really cool trick, and undoubtably helpful, but not general intelligence.

For once I’d like to hear an argument that doesn’t rely on Cartesian dualism, stoner metaphysics, or from people still clinging to GOFAI nostalgia like the Japanese holdouts.

Sure: I'll consider AI to be possibly sentient when it can tell me a thought it had 10 minutes ago, and then prove that it actually had that thought by showing me a peek into the workings of it's mind.

GPT is just text completion.

Back when I used to participate here actively, I ruminated a fair bit on why the Culture War Thread was so compelling to me. The urge to write, to argue, to contribute opinions, at full flow, was powerful to the point of absurdity, irrationality, compulsion. From an outside perspective, it made no sense. Everyone here is doubtless familiar with the "someone is wrong on the internet" meme, but why should it be so?

The best likeness I could come up with was bees building a hive. Pretty clearly, the bees have no conception of what they're doing or why, and yet they generate complex order. How? Instinct, clearly. They generate wax as part of their normal biological functions, and they put the wax where it should go. That this produces the hive that secures and sustains them is irrelevant to an individual drone; to the extent that they can be said to have "intentions", those intentions are simply to fullfil basic, granular biological imperatives.

It seemed to me that my own engagement with the Thread was analogous. When I read the thread, I was assessing my environment. If the environment seemed incorrect, if the wax was in the wrong place, I posted, moving the wax to the right place. Sometimes this process required deeper thought or analysis, and those moments were particularly interesting, but the majority of the time what I as engaged in was mainly memory and pattern-matching, call and response. @DaseindustriesLtd has mentioned a time or two how they find a new commenter, are impressed at first by their novel thinking, and then gradually come to see the repetitions and loops in their pattern of thought, till what seemed worth being excited about revealed itself as just another limited, too-human, simplistic pattern. I've definately had this experience with others. I've definately had it with myself.

All this to say, I think you should consider the degree to which "text completion" describes humans as well.

...Come to think of it, why do we do the Turing Test with a human and a computer? Why not have two computers talk to each other, while the human observes? What happens when one instance of ChatGPT talks to another?

What happens when one instance of ChatGPT talks to another?

Well, here's one example https://moritz.pm/posts/chatgpt-bing

What happens when one instance of ChatGPT talks to another?

They already did that...sort of, and with GPT-3.

I appreciate the response

Do you? What has it changed in your understanding of the situation, then?

but again, this isn't sentience. GPT is just text completion. It's not an intelligence, it's a tokenized list of words.

What makes your response superior to a product of text completion, then? Or really, text repetition. You remind me that transformers predict tokens. Okay. What does the word «just» add to your claim? You seem to believe that text completion is somehow insufficient for intelligence, but what is the actual argument for it? It is not self-evident.

I'll consider AI to be possibly sentient when it can tell me a thought it had 10 minutes ago, and then prove that it actually had that thought by showing me a peek into the workings of it's mind.

Wiki: «Sentience means having the capacity to have feelings.» Why do you talk of the AI sentience now when you have been dismissing its general intelligence just a few word before? Also, how isn't this just stoner metaphysics? Is your context window on the span of a few dozen tokens, or do you struggle to distinguish abstractions?

I am not saying this to put you down but, rather, to show that humans can be easily attacked on the same grounds as AIs.

Humans do not really have anything like general intelligence, we're just talking monkeys who know a few neat tricks, and if we concentrate real hard, we can emulate certain simple machines. We are surprisingly rigid and specialized things. The only reason we haven't yet made a superhuman AGI is precisely because we're that bad. It's not a high bar.

Agree. But this feels like a non-sequitur. AI won't need sentience to make the human race redundant.