The fundamental premise of this paper seems flawed -- take a measure specifically designed for the nuances of how human performance on a benchmark correlates with intelligence in the real world, and then pretend as if it makes sense to judge a machine's intelligence on that same basis, when machines do best on these kinds of benchmarks in a way that falls apart when it comes to the messiness of the real world.
This paper, for example, uses the 'dual N-back test' as part of its evaluation. In humans this relates to variation in our ability to use working memory, which in humans relates to 'g'; but it seems pretty meaningless when applied to transformers -- because the task itself has nothing intrinsically to do with intelligence, and of course 'dual N-back' should be easy for transformers -- they should have complete recall over their large context window.
Human intelligence tests are designed to measure variation in human intelligence -- it's silly to take those same isolated benchmarks and pretend they mean the same thing when applied to machines. Obviously a machine doing well on an IQ test doesn't mean that it will be able to do what a high IQ person could do in the messy real world; it's a benchmark, and it's only a meaningful benchmark because in humans IQ measures are designed to correlate with long-term outcomes and abilities.
That is, in humans, performance on these isolated benchmarks is correlated with our ability to exist in the messy real-world, but for AI, that correlation doesn't exist -- because the tests weren't designed to measure 'intelligence' per se, but human intelligence in the context of human lives.
The function of news is to help a democratic citizenry be critically informed, and that this kind of statistic doesn't accomplish what it set out to do, although it's certainly interesting for its own sake. I think it's a challenge of our age to figure out how to create institutions that are wise and don't simply bend to distorting pressures (money, politics, psychology).
For example, we do want terrorism over-represented relative to old-age-deaths. However, a responsible and self-aware media would really attempt to counteract 'availability bias' -- e.g. that due to the human mind what is repeated we tend to assume is actually more prevalent. But we don't have wise institutions at the moment.
The more general problem is that it is hard to quantitatively demonstrate the ways in which media fails at fulfilling its complex societal role, because it is a qualitative failure in general, although we can poke at it's edges for sure (e.g. fearmongering language probably has gone up, as has polarization on both sides of the aisle, and the amount of information-free 'babbling and speculating' in the immediate aftermath of some event has likely gone up over time).
Yeah -- I don't get why this is front-page -- reads like LLM quasi-insight:
"Through activation, lifeless equations became living systems. The neuron was no longer a mere calculator; it was a decider - a locus of transformation where signal met significance." -- wtf
The idealized (Science 1) / realpolitik (Science 2) dichotomy is both real and at first depressing. I also did a PhD in machine learning, and became quite disillusioned after seeing how the sausage was made, and how different the process is from how I had imagined it. At the same time -- engaging in 'game change' within Science 2 (perhaps not as a PhD, but after you have some security), is I think one of science's highest moral callings. The aim is not necessarily to inch Science 2 towards an impossible Science 1, but to help science to take itself more seriously (it really is a messy social process & there are ways that social process can work better or worse towards the public good -- itself a scientific question) -- and contribute towards science 2 becoming a better (and ideally better-at-self-improving) science 2.
I really think a lot of it falls on the funding agencies. While communication is key, there need to be some real questions answered. If a professor has 100 researchers under him and is churning out 2 papers per day how much is he really an expert? If his cousin is getting a PhD under him without doing the legwork shouldn't he be fired? Should we keep funding the same person excessively? It doesnt help that professors often rely on immigrant labor meaning that they can have a real choke hold over the lives of their staff. I really think there should be upper limits of what a professor can get away with, they need to be answerable to how tax payer money is being burned.
I think you’re pointing out a very important point; it seems like everyone pretends that you can ‘scale up’ scientific research, when it’s more like a service (that doesn’t scale). One problem might be that the people allocating the funding also stand to gain from pretending that they can “supervise” infinite amounts of research with no diminution of quality.
This is naive 'populism': There's no way to avoid 'allusion' writ large -- e.g. do you object to biblical references, or to references to particular experiences that only some people have (heartbreak, death of a father)? Sure, some communities basically 'write for themselves' in a way that becomes inaccessible to outsiders w/o a lot of work. But that's fine -- I like a McDonald's hamburger as well as really nuanced flavors (for whatever reason I like nuance in how I make oatmeal that likely few others probably appreciate). Film buffs like the nuance/allusions in that medium; etc. Your comment seems like: "The stuff I like is the best and does the most for humanity" -- I think there is indeed an argument for art that is broadly appreciable, but your comment is a form of the 'gatekeeping' you criticize -- it's gatekeeping for art that doesn't require a lot of effort (for you, and those like you) to appreciate.
> Despite the HN comments complaining about it being overwhelming and a dark reflection of how awful and distracting the internet is, clearly enough people enjoyed it to get to the front page.
Is this like a massive HN wooosh -- how can this be the top-voted comment?
From Neil Postman's 1985 "Amusing Ourselves to Death":
> “With television, we vault ourselves into a continuous, incoherent present.”
> “Spiritual devastation is more likely to come from an enemy with a smiling face.”
It's less about whether we "enjoy" the stimulation, more about what kind of people we become when we lose ourselves in this bizarre sea of superstimuli. We're like reinforcement agents creating adversarial examples for each other, drawing ourselves further out of any sort of meaningful life, into a fever dream where the most desirable job for the next generation is to be famous for being famous [1] rather than do anything for any kind of deeper purpose.
I like your description. I sometimes wonder if the final equilibrium state will be most people working on addictive products and the rest working on addiction treatment.
I'd say with the current state of things it's more like two singularities in which either:
- A landian-stephensian accelerationist timeline occurs where the majority of the urban population becomes some flavor of AGI-tuned VR junkie
- An extreme naturalistic counterculture movement occurs that causes majority of the civilized world to willingly roll themselves back 1 or 2 centuries technologically in order to feel something again
Perhaps the current obsession will just go the way of heavy drinking or smoking? Ie the population will eventually develop some partial immunity to the allure, but it won't even go away completely.
I personally believe that at some point, many people will realize that the majority of the people with economic means are the people who are able to concentrate and don’t waste all their time. Note that I don’t mean the super wealthy, I’m referring to people who are solidly middle class and have means. I know a lot of successful people who aren’t glued to their phones. I think there will be enough good and bad examples out there for people to start catching on.
- And there has to be the third, hyperminmaxers yearning forever more control and power trying to be(at) the machine. Thus becoming a reflection of the first.
- Fourth must be some sort of hybrid between denialist and creationist, whom I don't even want to envision through. Which would be a reflection of the second, but instead of withdrawing, they would bubble themselves into something terrifying version of the Amish.
I think it's kind of necessary to be exposed to ideas and views of people like Postman to even think of them when you play a game like this. The top comment is disappointing, but so it goes.
I enjoyed the game, for the attention to detail and making a mockery out of so many things in our daily lives that are in essence absurd, in such a brilliant yet simple way. The frowning Duolingo owl. The pillow delivery tracking made me chuckle. The only thing I missed was booking a an apartment on booking.com with a billion reminders to hurry up as the place might be gone any second, or doing an online check in. Although maybe it happened, I refreshed the game accidentally and never came back.
>> Despite the HN comments complaining about it being overwhelming and a dark reflection of how awful and distracting the internet is, clearly enough people enjoyed it to get to the front page.
> Is this like a massive HN wooosh -- how can this be the top-voted comment?
100% agree. I had to read that sentence and surrounding parts like 5 times to check if I was missing a satirical nod somewhere. It's like writing a review of Franz Kafka's books and saying "Despite what we may say about bureaucracy, clearly lots of people enjoy it because his books were best sellers!"
Lots of art is there to make you think, not to "enjoy" it.
I've been trying to disentangle myself from the internet for the last couple years. Maybe this hits better for those who haven't yet realized they are spending too much time online.
> High intake of sweetened beverages was associated with higher risk for most of the studied outcomes, for which positive linear associations were found. In contrast, a low intake of treats was associated with a higher risk of all the studied outcomes.
Not sure what to make of this -- some kind of other latent explanation (e.g. that many of those with the lowest intake of treats were on a diet due to bad health?).
From the discussion section:
> One aspect to take into consideration is however that there is a social tradition of “fika” in Sweden, where people get together with friends, relatives, or coworkers for coffee and pastries (41). Thus, one could hypothesize that the intake of treats is part of many people's everyday lives without necessarily being related with overall poor dietary or lifestyle patterns, and that it might be a marker of social life.
Isn't a simple possible explanation that people who are at high risk of (or already diagnosed with) these diseases make up a disproportionate chunk of the lowest sweets consumption bracket, simply because they're the ones who are actively restricting intake on advice from doctors?
I'd naively expect most healthy people to fall in the second-to-lowest bracket—they're not actively restricting, but they don't have the pervasive bad habits of the highest consumers.
That was my first thought as well, but the paper says that after excluding all people with a previous condition, that association remains unchanged. There might be something else, like particular diets, at play.
That would rule out pre-existing conditions but not rule out people who already knew they were high risk for reasons other than a previous diagnosis (such as blood work, family history, etc.).
Could it be that the type of sugar used in the soft drinks is different than the type of sugar used in baking and where the body processes them differently?
Yes, I'm questioning if high fructose corn syrup is worse for the body than table sugar. From everything I've ever seen, that tends to be the case.
Or could it be that the non-fructose corn syrup artificial sweeteners are having more of a negative impact as people do not limit their in-take of those beverages since there's "no sugars" or "0 calories"?
There might be a dietary difference, even though it seems superficially like there shouldn't be. This is my summary of the reaction from a chemist who did her master's degree in sugar chemistry: "Complex sugars (sucrose) and starches are broken up into glucose and fructose quite efficiently by enzymes in your digestive tract. Only glucose and fructose are absorbed into the bloodstream."
That chemist is my mom. So I'm biased. ;-)
The puzzle is whether table sugar and high fructose corn syrup have different dietary effects, and if they do, why. I don't think it's a settled matter.
There is clearly a difference in the uptake of fructose and glucose as you can see here, no insulin is required to uptake fructose. https://phdmuscle.com/fructose-metabolism/
If there is a difference in fructose and high fructose corn syrup it would likely be the fact that it is corn based. Corn has been selected to be farmed with high glycophosate (round up ready), which metabolically different from other corns and produces some proteins not found in non GM corn. https://enveurope.springeropen.com/articles/10.1186/s12302-0...
Its true, check the link and look at the metabolic pathway. insulin is a response to glycogen from the fructose 6 phosphate formed during glucose metabolism. since fructose converts to fructose 1 phosphate form no glycogen is released and so no insulin is required to uptake fructose.
Anyway, apparently HFCS contains about 50% glucose so it would require insulin, but less than pure glucose.
Fructose doesn't need insulin to be converted to glucose in the liver, but I don't think this is what you are thinking at all. Fructose isn't used by the body directly, it almost always has to be first converted to glucose first. Glucose does require insulin to be pulled into cells throughout the body. So fructose does require insulin to be used by the body for energy or to be stored into fat.
When you say insulin isn't needed for fructose it's a fundamental misunderstanding on a few different things.
My layman understanding is that insulin is used to absorb blood glucose into muscle and fat cells, while fructose is processed by the liver into other things (which may include glucose), but that initial processing doesn't require insulin.
Glucose gets broken up in the liver but fructose doesn't. So those with slow liver processing like Gilbert's syndrome the differences between the two make a bigger difference.
High fructose corn syrup floods the system (body) and it can't process it quick enough.
HFCS is high fructose compared to regular corn syrup. It's not necessarily high fructose compared to table sugar (sucrose). Table sugar is 50% fructose by weight; HFCS comes in 55% and 42% fructose varieties. The FDA says [1] the 42% variety, which has less fructose than table sugar, is used in most processed food other than soda. So fructose doesn't seem like a plausible mechanism for non-soda HFCS to be worse for you than table sugar.
I don't like HFCS--I think it gives sodas a flavor I can best describe as "sharp," based on blind taste tests I've done comparing same-brand sugar and HCFS sodas--but I don't see solid evidence that it's worse than sucrose. I think it's mostly used as a scapegoat by people who don't want to conclude that sugar in general is bad for you.
> The puzzle is whether table sugar and high fructose corn syrup have different dietary effects, and if they do, why. I don't think it's a settled matter.
I don't know how you would design a study around it, and this is an anecdote, but I find sugar sweetened soda to be much less palletable than HFCS sweetened soda.
That wouldn't necessarily show up in a study comparing equal intake, but could be a factor in reducing intake of sweetened beverages in places without HFCS compared to those places with.
I'll go one step further. A gram of sugar in an orange is not the same as a gram of sugar in a teaspoon.
Medium matters - that sugar in an orange is dissolved in the juice which is locked in the pulp.
Who knows what the bioavailability of the sugar is in that messy fibrous orange you partially chewed? I guarantee it'll be less than that pure refined sugar in the spoon.
Our "datafication" of nutrition and stuff like calories has led to so much silly pseudoscience.
Calories are my biggest pet peeve of BS to take with a grain of salt. Why are we basing our nutrition and diet on the performance of the food in a bomb calorimeter?
Table sugar is 1/2 glucose and half fructose. An orange is 100% fructose. The medium of the pulp allows for slower absorption and the fiber does a number of things like simulating the intestinal walls.
In fact an orange is a mixture of glucose, fructose, and sucrose (which itself is glucose and fructose). An orange actually looks surprisingly like table sugar in terms of the sugar types, although you are correct that the pulp / fiber / etc do mean that the response of our body is likely different.
It is known that people who consume sugary drinks are at higher risk of death from disgestive diseases, which is less so for people who consume artificially-sweetened drinks. So there may be a causal link.
I think that at least it is not obvious that sucrose intake should be equivalent to invert sugar intake. In order for sucrose to be saccharified, the enzyme sucrase must be produced; in order for sucrase to be produced, some internal regulation must occur; if some internal regulation occurs, other effects are at least possible. But no such effect has been clearly demonstrated as far as I can tell.
>Because of its low cost and long shelf-life, HFCS is used widely in manufacturing many food products, including candy, throughout the United States. However, due to strict EU regulations, HFCS is banned in much of Europe, including Sweden.
I think the evidence merely suggests that his DEA will be ineffective. His initial pick for head of the DEA, Florida sheriff Chad Chronister, went two days [0] from being announced to be being "pulled out" on account of his lack of experience with the border and his arrest of a megachurch pastor for COVID public gathering rule violation [1].
My assumption is that the GP expects that it will be gutted by the Department of Government Efficiency and/or be significantly laid off via Schedule F conversion [2]. However, rules made earlier this year by the OPM should limit that specific method of getting rid of civil servants [3].
> I think high fructose corn syrup may be a North American thing
It is. It's not a common ingredient here in Europe or most places in the world. Corn syrup is a weird side effect of subsidized corn farming in the US. This causes all sorts of health issues in the population.
Anyway, you'd struggle to find it in supermarkets or in food outside the US. Even coca cola doesn't use corn syrup in it's beverages outside the US. As far as I know, there is no major taste difference and I've never heard any US people complain about their coke tasting a bit off outside the US. But I don't drink cola that often myself.
> As far as I know, there is no major taste difference and I've never heard any US people complain about their coke tasting a bit off outside the US.
Apparently, you are not well informed about the happenings across the pond. It is well known that Coke made in Mexico uses sugar, and it is a known thing in the US to get Mexican Coke. There used to be a small plant in Dublin, Texas that made Dr. Pepper with sugar that was a well known for the different taste. Eventually Dr. Pepper won it's case against the Dublin plant and it is no longer available.
To claim that there's no taste difference between sugar vs HCFS just shows you're not well informed on the topic.
> Regardless of what was actually in the serving containers, people stuck by their original choice. Those who preferred what really was the Mexican Coke the first time (we'll call these guys the Mexico Boosters) unanimously picked the Coke that I told them was the Mexican Coke the second time, whether it really was or was not. Even when the containers were completely removed from the test and the Coke was served in plastic cups, the Coke labeled as Mexican was picked by the Mexico Boosters every time.
Dr. Pepper does make, in limited quantities, its own version[1] of the real sugar deal. It isn't stocked at major supermarkets - in fact, I've only ever bought it at a liquor store in Texas. It's fantastic!
US people who have expressed opinions on the matter near me claim to like the taste of Coca Cola from other countries better. Specifically Mexican Coke, though I've heard similar sentiments with Canadian Coke.
We are just making hypothesis here but I doubt its that. HFCS has been looked at a lot and while there are amounts of people that think it can make them sick, I don't believe in normal doses it does harm in the base case.
The easier explanation is that most "treats" are low glycemic as they are paired with a fat. Where with a can of coke you are ingesting 40g of sugar mixed with some flavorings and water.
There are some people that argue this actually. Dr Fungs book on type 2 diabetes goes into it.
Basically his take is that while glucose can be distributed in the bloodstream, fructose is broken down by the liver. With high intakes of fructose comes an overwhelming of the liver and higher incidence of fatty liver plus visceral fat. And those things are highly associated with bad health outcomes
> that many of those with the lowest intake of treats were on a diet due to bad health
I haven't read this study, but any study that isn't complete garbage normally would examine the to groups across all variables (age, gender, weight, health, etc) and deliberately control for any systematic differences between groups.
You can never control for all variables when humans are involved. (You can only control for variables that your data contains.)
EDIT: From the article: "[There's a] need for future research to incorporate more accurate measures, such as urinary sodium excretion, to more precisely assess the impact of sodium intake as a potential confounding factor."
Simple. Metaphysics is not math. Ethics is not math. Really, the only intersection is formal logic (until a certain German/Austrian mathematician blew it all up with his annoying theorems)
But applied mathematics can have ethical impact -- e.g. the concept of whether a human should trust the output of a particular language model. So GP's idea of 'trust' not applying because an object has its basis in math seems like a false dividing line. Ultimately everything can be grounded in things such as math as far as we know, although its not useful to reason about e.g. ethics from thinking about the mathematics of neuronal behavior.
This is not true. Lots of things have no mathematical foundations because it is impossible to state them formally/symbolically. If you can not specify it formally then it is not mathematics. AI is mathematics because software/code/hardware is mathematics so all the hullabaloo about "safety" makes absolutely no sense other than as a marketing gimmick. Even alignment has been co-opted by OpenAI's marketing department to sell more subscriptions.
But in any event, the endgame of AI is a machine god that perpetuates itself and keeps humans around as pets. That is the best case scenario because by most measures the developed world is already a mechanical apparatus and the only missing piece for its perpetuation is the mechanical brain.
As usual, I can build this mechanical brain for $80B so tell your VC friends.
I don't get this line of logic -- of course software has safety implications, because people use it for things in the real world. It isn't "math' that is cleanly separable from the rest of humanity; its training data comes from humanity, and it will be used towards human goals. AI is entangled with the rest of human dealings.
Whether AI poses existential threats for us or not, I'm open to either direction, but that the experts (e.g. Hinton, LeCun) are divided is reason enough to be concerned.
The way safety is handled in real world situations is through legal and monetary incentives. If the tanker you are driving to the gas station blows up then people get fired (no pun intended) and face legal repercussions. This is the case for anything that must operate in the real world. Safety is defined and then legally enforced. AI safety is no different, if an AI system makes a mistake then the operators of that system must be held liable. That's it, everything else about extinction and other sci-fi plots has no bearing on how these systems should be deployed and managed.
I have no idea what people talk about when they say LLMs must be safe. It generates words, what exactly about words is unsafe?
The long-term impact of this paper has confused me from a technical lens, although I get it from a political lens. I'm glad it brings up the risks from LLMs but makes technical/philosophical claims which seemed poorly supported and empirically have not held up -- imo because they chose not to engage with RLHF at all (which was deployed through GPT-3 at the time; and enables grounding + getting around 'parrotness'), and uses over-the-top language ("stochastic parrot") which seems very poorly to capture what it feels like to meaningfully engage with e.g. models like GPT-4.
This paper, for example, uses the 'dual N-back test' as part of its evaluation. In humans this relates to variation in our ability to use working memory, which in humans relates to 'g'; but it seems pretty meaningless when applied to transformers -- because the task itself has nothing intrinsically to do with intelligence, and of course 'dual N-back' should be easy for transformers -- they should have complete recall over their large context window.
Human intelligence tests are designed to measure variation in human intelligence -- it's silly to take those same isolated benchmarks and pretend they mean the same thing when applied to machines. Obviously a machine doing well on an IQ test doesn't mean that it will be able to do what a high IQ person could do in the messy real world; it's a benchmark, and it's only a meaningful benchmark because in humans IQ measures are designed to correlate with long-term outcomes and abilities.
That is, in humans, performance on these isolated benchmarks is correlated with our ability to exist in the messy real-world, but for AI, that correlation doesn't exist -- because the tests weren't designed to measure 'intelligence' per se, but human intelligence in the context of human lives.