Comment by tomaytotomato
5 months ago
LLMs cannot think, they just guess with some magic with the help of things like softmax (https://en.wikipedia.org/wiki/Softmax_function), that makes it seem intelligent.
We aren't enslaved or at any risk just yet with this form of "thinking machine", but we do need to culutrally and emotionally come to terms with them.
Let's not have any other "jihad" just yet, there's plenty of other jihads going on right now.
Note that the article doesn’t say that AI itself would do the enslavement; on the contrary:
> With the rise of AI came the rise of a technocratic class that created and controlled these machines, leading to oppressive structures over knowledge and the economy.
It also quotes Dune, where it is said more directly:
> Once men turned their thinking over to machines in the hope that this would set them free. But that only permitted other men with machines to enslave them.
I really despise when the term technocrat is misused. It has a very specific meaning in politics and is nowhere near "broligarch", "plutocrat from the tech industry", or any of the thousand different, legitimate ways to describe what people sometimes mean when they mistakenly say "technocrat". How can a political opinion be taken seriously when this extremely basic level of literacy on the topic is missing?
Not to "well ackchyually" your "well ackchyually", but language evolves. I guess in a few years we'll consider it just another instance of semantic shift, perhaps with a splash of folk etymology[1].
[1]: https://en.wikipedia.org/wiki/Folk_etymology
3 replies →
It is difficult to seriously consider the opinion of someone who clearly fails to understand the difference between denotative and connotative meaning.
When "guess with some magic" can solve long-standing problems in mathematics that no human had been able to, it seems fair to ask whether it's approaching risky levels of intelligence.
I mean we've been using technology to solve long-standing problems in mathematics that no human had been able to do for years. Is Deep Blue sentient because it can beat any human at chess?
Deep Blue's strength was leveraging massive compute to execute a task-specific, human-written algorithm. The problems which LLMs are tackling don't elude mathematicians because they require too much number-crunching, but because they demand creative problem solving. The latter seems more profound, even if it doesn't imply sentience.
1 reply →
[dead]
> We aren't enslaved
The people complaining about becoming stupider because of their AI usage might have something to say. Those that wish they didn't have to use AI but are coerced by their workplace might disagree. What of the day something negative happens to you based on the wrong, unreviewed decision without appeal by an AI agent? Do you truly believe this day will never come?
You don't need literal chains to be a slave.
Man it annoys me to no end that this basic fact is so under discussed.
It’s as if calling something AI is all that’s needed to justify it being the same thing as Skynet, Dune automation, Terminators, Neuromancer, and every other cultural trope using the label of AI.
The ultimate fact is that LLMs are a pretty powerful technology that has little to nothing to do with most fictional depictions of intelligent machines.
Sloppy thinking all the way down.
Yesterday's plane making a turnaround because of a Bluetooth speaker with the name "bomb" (media says "a 4 letter word") makes me think the Skynet-level destruction is a risk, because humans are already too chicken-shit to take responsibility for decisions. They'd rather feel safe by just following the procedure, however stupid it is. And in the the future the procudre might be "Do whatever the AI says".
My thought about the turnaround: https://news.ycombinator.com/item?id=48343390
Absolutely. The real danger is that people treat LLMs like actually intelligent entities and stop making decisions themselves. Not that the LLMs will be conscious and evil in the way Sci-fi portrays.
People worry about what will come in the future even if today's tech has limits.
We don't have "Skynet, Dune automation, Terminators, Neuromancer, and every other cultural trope using the label of AI" today simply because LLM technology doesn't seem to directly lead to those, but the end goal of AI research is the creation of those.
It annoys me to no end that people keep looking at the finger while pointing at the moon. No, we don't have Skynet today, but you can believe Sam Altman would have loved to see Skynet appear out of ChatGPT. That is the end goal, and it's worth discussing now, not when it's too late.
We’ll have to agree to disagree then. Personally I think treating sci-fi tropes as likely outcomes is a major distraction from the actual real likely and near term outcomes from LLMs. There are enough real problems there, we don’t need to indulge everyone’s fictional scenarios too.
1 reply →