Damin Kim

@2196F3 @Qivshi1 Yet the human brain does it without any effort.

@Qivshi1 A lora large enough yo remember name will erase the learned concepts everywhere else pretty quickly

The best way to do LLM memory would be through something like a big LORA that continuously updates weights eg to learn a person's name well, you have to move beyond a little note (how LLM memory is currently done) and just learn the association automatically (from the weights)

@andrewgwils Big Data. Machine Learning. Artificial Intelligence.

It's funny how phrases like "generative AI" and "big data" are an efficient way of signaling that you are not a domain expert.

@FilmLadd But imagine how cool it would be

The US Military ultimately determined that the "Rods from God" concept wouldn't work as initially imagined. All sorts of issues, but the primary one: it results in an energy release no more powerful than a conventional bomb.

@Joy678320811181 @Shiyanying @fchollet I am obviously so much better for not even expressing my opinion and just critiquing you

@adroptable @Shiyanying @fchollet Very high. Very mighty sir. Great debate. Way to really get to the heart of the issue and avoid getting caught in the weeds. Well done.

One of the biggest misconceptions people have about intelligence is seeing it as some kind of unbounded scalar stat, like height. "Future AI will have 10,000 IQ", that sort of thing. Intelligence is a conversion ratio, with an optimality bound. Increasing intelligence is not so much like "making the tower taller", it's more like "making the ball rounder". At some point it's already pretty damn spherical and any improvement is marginal. Now of course smart humans aren't quite at the optimal bound yet on an individual level, and machines will have many advantages besides intelligence -- mostly the removal of biological bottlenecks: greater processing speed, unlimited working memory, unlimited memory with perfect recall... but these are mostly things humans can also access through externalized cognitive tools.

You did not really have an objective point. You just made a bunch of statements. It's at the very best an opinion piece. You keep saying things with conviction as though you have some divine source of information. "everyone understood". I get it though. You disagree. No real meaning. Just pure conflict of belief. Same as OPs statement. Meaningless asspull. You are just easy to put down since you use concepts without understanding their meaning.

@adroptable @Shiyanying @fchollet I understand the point. But everyone understood my point also.

@redaction there might be some innacuracies due to the fact that the first graph uses adjusted dollars and the second one includes benefits

@redaction The data doesn't corroborate this unfortunately. ~ 50 % of people are in the first 2 points of the graph. While less than ~ 1 % are in the last 3.

Modernity is genius because it’s essentially a sorting algorithm where the losers viciously defend their own right to be sorted against In the absence of traditional guard rails, they are permitted to self-destruct openly, weeding themselves out of the gene pool

@menhguin @fchollet the question is : is there a complex problem one von neumann can't solve that more than one von neumann can solve? (excluding volume limits)

@fchollet no, i don't think this analogy makes sense. there's no specific reason why humans are anywhere near the theoretical optimality bound. there is no reason von neumann can't be the species average, and then another von-neumann level genius comes along in that population.

@Joy678320811181 @Shiyanying @fchollet IQ is the quotient of general intelligence - G. It is measured relative to a norm made of other people. You already used the term 500 IQ which implies that our tests are capable of capturing that intelligence. He just pointed out that that is not the correct test to use.

@Shiyanying @fchollet “We do not have a test for something therefore it cannot exist” brilliant.

@JohnDiddy9 @AlperKaanAnkara @ItsTaz1989 Yeah, he was probably referring to overall technology. Egypt probably had better non military technology. But having superior military technology trumps all.

@adroptable @AlperKaanAnkara @ItsTaz1989 Yea I’m not sure why he even believed that they were technologically inferior in the first place. It’s clear they were not.

There’s this stupid idea which just won’t die: Cultures which were technologically & economically inferior to those that colonised/defeated/enslaved them were somehow morally superior, despite the fact that the only thing that prevented them doing the same was that inferiority.

Even sillier that calling someone silly and bringing nothing to the conversation ? My point is that global warming doesn't actually "hurt" plants. It hurts people. The zeitgeist is pretending that CO2 is somehow "bad" for the planet and plant life when in actuality is at worst a self correcting mechanism. PFCs were something that was really bad for every living thing and humanity got rid of them (for the most part).

@adroptable @ChrisMartzWX Such a silly thing to say. Plants use CO2 - as if there was anyone who didn't know.

China isn't turning into a “green superpower.” Any renewable energy systems they're installing only add to existing energy sources; they are not replacing fossil fuels at all. China is increasing their use of ALL energy.

@AlperKaanAnkara @ItsTaz1989 well depends how you define technological superiority or inferiority. imo if you win you have the superior military technology regardless of it's sophistication.

@ItsTaz1989 For many centuries Egyptians were the most advanced society on Earth and they didn't show much inclination towards conquest. They only built trade posts here and there, in Sudan, Somalia, etc. Then they were conquered by Hyksos nomads -a technologically inferior society.

That is my whole point. Also you don't understand that private universities with low absolute attendance numbers are not the same thing as public universities where a bell curve is relevant. If you have 20 people in a class from selected backgrounds you are not getting a bell curve.

If everyone in your classes are getting exactly the same grade then yes a strict bell curve is needed because the teacher is either too nice or too lazy to differentiate the students A bell curve forces them too. Everyone in the class doesn’t have the same mastery of the material. If it appears they do, the test of poorly designed

Why don't we simply have an Elo system for grades, where you receive more points for getting a high relative grade in classes with other people with high GPAs (and vice versa)? This actually seems not terribly hard to implement, compared to what Harvard seems to be trying.

The reason why Harvard wants to change the grading system is because everyone is graduating with the same (A) grades and grades have lost the competence signal. My anecdotal experience has nothing to do with this. But, yes I've been in plenty of classes where everyone got exactly the same grade. Also, if you select from people that roughly have the same grades going in, maybe they will have the same grades going out. Are you incapable of productive conversation? Yes ? That's what i thought.

@adroptable @BenShindel Have you ever been in a class where everyone got exactly the same grade? No, ya that’s what I thought.

@erat_perfect @BenShindel What if everyone gets the same grade ? They should just focus on making better tests, not on grading them better.

@BenShindel They should. Harvard shouldn’t stop being competitive once you get in. You want to play with the big boys then step up. Your attitude is what has led to this rampant grade inflation at Harvard.

@eigenrobot All animals are mentally retarded compared to humans. You don't need to be able to be sharp to do what most animals do. Some of them are actually selected for less intelligence.

just realized you never see nature documentaries about mentally retarded animals but I bet many exist and play important roles in their ecosystems

@gentlegossip I'm sure they know what they look like. Maybe they don't know if they look good or bad in the eye of the average beholder.

MedGold was so real for saying women don't know what they look like

@JoeECusick @FunctorFact There is time as the abstraction and time as the underlying principle. The abstraction layer that is natural language doesn't allow us to differentiate between the two.

@FunctorFact Time is more than an abstraction, it runs concurrent with motion.

'The purpose of abstraction is not to be vague, but to create a new semantic level in which one can be absolutely precise' - Edsger Dijkstra

@cremieuxrecueil Ok , but like did they calculate how much it would spread? Even if you treat half the head, if the effective reach of the therapy is to your toes it doesn't really matter does it?

I found him! One half of his head was just treated and the other half had a placebo applied. I don't know which is which. Any guesses?