I think it does accurately model the part of the brain that forms predictions from observations—including predictions about what a speaker is going to say next, which lets human listeners focus on the surprising/informative parts. But with LLMs they just keep feeding it its own output as if it were a third party whose next words it’s trying to predict.

It’s like a child describing an imaginary friend, if you keep repeating “And what does your friend say after that?”

Fish.

My apartment building turns 100 this year.

There are three factors that might prevent you from using an online platform to share your opinion:

• The government might impose content restrictions on the platform

• The owners (or delegated moderators) of the platform might impose restrictions on its users

• The users of the platform might use its tools to police the opinions of other users.

“Freedom of speech” normally just applies to the first. Is that what you mean when you say you’re not “allowed” to speak freely?

Dont say Prometheus

Ok… it was Προμηθεύς.

These days elders are printing screenshots of emails (and wondering why they’re cut off).

IMO the focus should have always been on the potential for AI to produce copyright-violating output, not on the method of training.

Why would the article’s credited authors pass up the chance to improve their own health status and health satisfaction?

Critical paragraph:

Our research highlights the importance of Germany’s unique institutional context, characterized by strong labor protections, extensive union representation, and comprehensive employment legislation. These factors, combined with Germany’s gradual adoption of AI technologies, create an environment where AI is more likely to complement rather than displace worker skills, mitigating some of the negative labor market effects observed in countries like the US.

How much of their profits do the Waltons put back into the Bentonville economy?

What percentage of the money you spend in that Walmart will stay in Bentonville, versus money you spend in alternative local shops?

If you think they’d be open to it, try Bayes’ theorem. Ask them to give percent likelihoods for the following:

A. The odds that the government (or whoever) is trying to kill everyone, before taking the evidence of excess deaths into account
B. The odds of seeing excess deaths for any possible reason, not just their conspiracy hypothesis
C. The odds of seeing excess deaths if the conspiracy hypothesis were true.

Then logically, the odds of the conspiracy being real given the excess deaths should be A*C/B. If you disagree with them on the outcome, you must disagree on one or more of the assumptions (probably A—if it’s B, you can find the objective odds by checking historical data).

If you still disagree on the prior assumption (A), you can set aside the excess deaths argument and ask what other evidence led them to form that prior assumption. Then you can repeat the process until you either reach agreement or they’re left with an assumption they have no evidence for.

That makes sense—being raised by ChatGPT might be marginally better than being raised by Sam Altman.

Thanks! I hate it.

They’d have to prove they created the god in question first, which would undermine every other claim they made for it.

Words aren’t isomorphic to their dictionary definitions—words had commonly-accepted meanings long before the existence of dictionaries. Dictionary definitions are just an attempt to come up with a heuristic for identifying things as instances of the term in question, but they’re never perfect—and the real-world usage is ontologically prior.

If the dictionary definition of sandwich fails to distinguish cakes from sandwiches, it’s just an imperfect definition (like all definitions are)—and we can leave it at that.

If life is an experiment, and there is a measuring tool

So that’s what keeps causing our wave function to collapse!

How does that compare to the growth in size of the overall code base?

They’ve got to do something with the cows once they’re too old to milk.

Sure… but everything that’s been trending steadily upward over the last 50 years correlates with everything else that’s been doing the same.

To suggest causation, we’d need to find differences in rates of sunscreen use between locations, and see if that correlates with local differences in depression rates (after controlling for other confounding factors, etc).

I think it’s more likely that you need to be able to empathize with someone in order to recognize when they’re being empathetic, and you empathize most with the middle class.