Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

IQ tests results for AI

trackingai.org

Post reply on HN
Print view View original

221–230 of 359 posts

Re: IQ tests results for AI

#221

Post by CamperBob2 »

Telemakhos wrote: post #195
CamperBob2 wrote: post #108

Earlier quoted context omitted.

The irony isn't that the GP is right, because they are. "AI does not understand anything: it just predicts the next token based on previous tokens and statistics." The irony is that we've recently learned that "just predicting the next token" is good enough to hack code, compose music and poetry, write stories, win math competitions -- and yes, "give synonyms matching a particular metrical pattern" (good luck composi…

AI fails miserably at writing poetry in Latin and Greek, precisely because it cannot apply metrical rules. Predicting the next token does not produce correct verse. Perhaps it works for stress-based songs in English, but it does not in languages with quantitative (moraic) meter, nor can AI scan meter correctly. Schoolchildren can and do: it's a well understood domain with simple rules, if you can apply rules. Token p…

How'd you do at that, before someone taught you?

If someone cared enough to train a model on Latin and Greek theory, then rest assured it would do just fine. It'd just be a waste of weights from the perspective of almost everyone else, though.

Re: IQ tests results for AI

#222

Post by clove »

matthewdgreen wrote: post #78
nsoonhui wrote: post #31

Earlier quoted context omitted.

Another component of this theory concerning g is that it's largely genetic, and immune to "intervention" AKA stability as you mentioned. See the classic "The Bell Curve" for a full exposition. Which makes me wonder what's the point of all the intervention in the form of teaching/parenting styles and whatnot, if g factor is nature and immutable by large? What's the logic of the educators here?

If IQ potential was 50% genetic, then the teaching would potentially raise your actual IQ by affecting the other 50% which is huge. IQs scores in populations and individuals change based on education, nutrition, etc. But even if we hypothesized a pretend world where “g” was magically 100% genetic, this (imaginary) measure is just potential. It is not true that an uneducated, untrained person will be able to perform t…

You've made the mistake of thinking that if IQ were 50% genetic (it isn't - it's way more than that, but that's beside the point), then the remaining proportion is completely (non-shared) environmental.

Researchers in this field actually break down the non-genetic component into four major components: 1. Shared environment 2. Non-shared environment 3. Error

Shared environment accounts for so little variance that it might as well be ignored; while non-shared accounts for little more than error.

Note that including error in the non-genetic component, just as you've done in your post above, you are viscerally downplaying the otherwise undeniably predictive link from genes to IQ. In other words, whatever number you give is automatically deflated due to the way a psychometric is measured.

This has never been the source of debate. Back when I was going through grad school in intelligence, people didn't have to overthink how they presented the data. Intelligence was already a mature field, and we discussed the data openly. But in the past couple decades or so, a lot of people such as yourself popped up, attempting to craft irrelevant, statistically incorrect arguments against the results of certain well-established psychometrics that happen to not fit within whatever mental world your brand of politics ascribes.

If you really cared about the data, you'd be discussing the numbers. But your interest in this previously niche topic isn't in understanding reality; it's in justifying your worldview, which is why you deny the established data, immediately present a caveat stating that the data doesn't matter in the first place, appeal to emotions, and finish it all off by claiming those who disagree with you have been brainwashed. None of those four arguments have any merit in a genuine discussion on this topic.

Re: IQ tests results for AI

#223

Post by judofyr »

lumb63 wrote: post #197
judofyr wrote: post #99

Earlier quoted context omitted.

Your first link (Wikipedia) directly contradicts your examples: > Although IQ differences between individuals have been shown to have a large hereditary component, it does not follow that disparities in IQ between groups have a genetic basis[18][19][20][21]. The scientific consensus is that genetics does not explain average differences in IQ test performance between racial groups.[22][23][24][25][26][27].

I suspect that this is an instance where “the scientific consensus” is wrong because to suggest contrary to that is wrongthink and enough to have one ostracized not only from science, but also society as a whole. I would love to be wrong, so if someone could explain this to me, I’d be very receptive to an explanation of why this logic is wrong: First, let’s substitute emotionally charged terms for more neutral terms;…

> Suggest contrary to that is wrongthink and enough to have one ostracized not only from science, but also society as a whole.

There's many scientists who have published the "contrary". They were not ostracized from science or from society as a whole. These saw next to none negative impact to their position while they were alive. Other scientists have published rebuttals and later some of the originals articles have been retracted.

J. Philippe Rushton: 250 published articles, 6 books, the most famous university professor in Canada. Retractions of this work came 8 years after his death.

Arthur Jensen: Wrote a controversial paper in 1969. Ended up publishing 400 articles. Remained a professor for his full life.

Hans Eysenck: The most cited living psychologist in peer-reviewed scientific journal literature. It took more than 20 years before any of his papers were retracted.

There's a lot of published articles about the "contrary view" that you can read. You can also read the rebuttals by the current scientific consensus (cited above).

> The analogous claim would therefore be that “although height differences have a large hereditary component, it does not follow that disparities in height between families have a genetic basis.” This seems very clearly false to me.

But this is not an analogous claim since you're talking about disparities between families. The analogous claim would be: "although height differences have a large hereditary component, it does not follow that disparities in height between groups have a genetic basis".

A very simple example for height[1]: The Japanese grew 10 cm taller from mid-20th century to early 2000s. Originally people thought that the shortness of the Japanese was related to their genetics, but this rapid growth (which also correlates with their improved economy) suggests that the group difference between Japanese and other groups was not related to the genetic component of height variance.

[1]: Secular Changes in Relative Height of Children in Japan, South Korea and Taiwan: Is “Genetics” the Key Determinant? https://biomedgrid.com/pdf/AJBSR.MS.ID.000857.pdf

Re: IQ tests results for AI

#224

Post by jacquesm »

lumb63 wrote: post #197
judofyr wrote: post #99

Earlier quoted context omitted.

Your first link (Wikipedia) directly contradicts your examples: > Although IQ differences between individuals have been shown to have a large hereditary component, it does not follow that disparities in IQ between groups have a genetic basis[18][19][20][21]. The scientific consensus is that genetics does not explain average differences in IQ test performance between racial groups.[22][23][24][25][26][27].

I suspect that this is an instance where “the scientific consensus” is wrong because to suggest contrary to that is wrongthink and enough to have one ostracized not only from science, but also society as a whole. I would love to be wrong, so if someone could explain this to me, I’d be very receptive to an explanation of why this logic is wrong: First, let’s substitute emotionally charged terms for more neutral terms;…

You're wrong. Some of the smartest kids I know are from immigrant children. It is their background - and society's response to that background - that hinders them, not their genetics. More so if they aren't lily white. Note how anything you say about this subject will be used to generalize to much larger groups (of which you can find some prime examples in this very thread) than the ones that IQ tests themselves target: individuals. And you can't say much about how an individual scores on their IQ test without accounting for their environment because that's a massive factor.

All of your arguments more or less equate to 'I don't understand the subject matter, but I'd like to see my biases confirmed'. And, predictably, you see your biases confirmed. But some of the smartest individuals that ever lived came from backgrounds and populations that - assuming the genetic component is as strong as you make it out to be - would have precluded them from being that smart.

Bluntly: wealth and access to opportunity have as much to do with how well you score on an IQ test versus what your genetic make-up is. Yes, it is a factor. No, it is not such a massive factor that it dwarfs out the other two once you start looking at larger groups. Income disparity and nutrition alone already negate it.

https://en.wikipedia.org/wiki/Effect_of_health_on_intelligen...

And that's just looking at that particular individual, good luck to you if your mom and dad were highly intelligent but you ended up as the child of drugs or alcohol consumers. Nothing you personally can do about that is going to make up for that difference vs growing up as the child of affluent and moderately intelligent people.

IQ tests are a very imprecise yardstick, and drawing far reaching conclusions about the results without appreciating the complexity behind squashing a multi-dimensional question into a single scalar, especially when you are starting out from a very biased position is not going to lead to a happy ending. Before you know it you'll be measuring skull volume.

Re: IQ tests results for AI

#225

Post by api »

somenameforme wrote: post #205
api wrote: post #155

Earlier quoted context omitted.

Throughout 99% of human history, the economy was mostly zero sum. If someone was rich it was because they stole it. If a group was wealthier it was because they stole it. By "stole it" I'm including theft of labor through slavery as well as the usual conquests and raiding. There was little to no innovation over time spans as long as thousands of years. Most of human culture and philosophy evolved during these periods…

You're ignoring inflation. Real (inflation adjusted) median earnings are up 12%, in total, over the past 45 years. [1] That also understates the problem because many things (education, healthcare, debt costs, etc) are understated or not even considered in inflation measurements. There's also been a substantial increase in baseline necessities for the average person (internet, computing device, etc). Factor these in a…

So the change from sleeping on hay and riding horses and subsistence farming to this is just a facade?

I'm not talking about numeric wealth, which I agree is hand wavey, but actual tangible physical wealth as well as our insane increase in knowledge and capability.

I can go watch videos from the surface of Mars and pictures of galaxies from the beginning of time, then go to the doctor and get injected with something that programs my immune system to resist diseases I've never encountered, then ask an AI to explain any concept from the history of science or mathematics. I am middle class and live better than a pharaoh in a lot of ways, and some of the things I just described are available to the very poor. This is simply nuts and it is not a facade or an illusion.

The collapse of the numeric financial industry won't take all this away unless we let it.

Re: IQ tests results for AI

#226

Post by LNSY »

jedberg wrote: post #220

The more interesting link (to me) is this one: https://www.trackingai.org/political-test They run each model through the political leaning quiz. Spoiler alert: They all fall into the Left/Liberal box. Even Grok. Which I guess I already knew but still find interesting.

There's a way to fix this political bias: feed it a bunch of bad code

https://www.quantamagazine.org/the-ai-was-fed-sloppy-code-it...

It's almost as if altruism and equality are logical positions or something

Re: IQ tests results for AI

#227

Post by nialse »

bearl wrote: post #158
nialse wrote: post #85

Earlier quoted context omitted.

Just to clarify: the prevailing notion in many context were that genetics does not matter and thus given the necessary social and educational interventions every human would prosper. Sadly, this is not the case. We are limited by our biology AND the extend we and our environment manages us to fulfill our potential.

But we are not limited by our biology. With tools and technology we can change our limits. From pharmaceutical tools like adderal to neurolink style brain implants to ai assistants to genetic engineering, the limitations in our cognitive capacity are becoming less salient every day. The importance of g in the future asymptotically approaches 0 the farther out you go, at least in terms of economic outcomes. It will al…

“It [g] will always be important for moral reasoning as I’m frequently reminded.”

Excellent quote! Unfortunately not all high g people engage in moral reasoning, and I fear that they will tend to exploit lower g people, rather than to help them utilize AI to compensate. There is a real opportunity to help individuals with cognitive impairments enhance their abilities with AI. The question is how, and how they collectively feel about it.

Re: IQ tests results for AI

#228

Post by jedberg »

ahmedhawas123 wrote: post #210

Would be great to add a few human benchmarks on this (e.g., average US IQ, Ivy League average, human 80th percentile). Also understanding some IQ per cost metric could be fun. Overall this is fun but not sure anyone in their right mind will be selecting an LLM based on this IQ benchmark

The IQ curve is designed so that 100 is average. So two of your questions can be answered with math:

> average US IQ

100

>human 80th percentile

113

The last one, Ivy League average, can be guessed based on published data. The median SAT score of an Ivy League attendee is 1500. A 1500 on the SAT is roughly an IQ of 130-140.

So in theory, the median Ivy League attendee is at the genius level.

Re: IQ tests results for AI

#229

Post by nialse »

breakyerself wrote: post #120
nialse wrote: post #63

Earlier quoted context omitted.

Do note that The Bell Curve is not considered controversial in general. The part about race and genetics is. Also genes being the sole predictor of IQ is not an accurate description of the book’s premise.

No the whole book is controversial. It's a political argument for dismantling the welfare state disguised as a review of science. They laundered a bunch of work by racial eugenicists along with a bunch of other junk science methodology. https://youtu.be/UBc7qBS1Ujo?si=bzVMwGU4XjPrk4sr

I hear you are invested in this line of thought, and that is okay. I just don’t agree with the analysis, and not with the labeling.
Post reply on HN