And BTW, to all the people obssessed with "paleo" / "cave men" diets and think that means "meat":

While there certainly were (just like today) some hunter-gatherer societies (such as in the high arctic) that got most of their calories from meat, the average hunter-gatherer got the majority of their calories from plants. But not highly bred high-calorie plants, but rather high-fibre wild plants, meaning that they had to eat significantly more of these fibre-loaded plants. Your average hunter-gatherer ancestor ate way more fibre than a person eating a "modern" diet today eats.

A proper "paleo" / "cave man" diet isn't a high-meat diet - it's a high fibre diet.

No, the last fad was antioxidants. Bodybuilders being obsessed with protein is entirely different from maxing protein in normal consumer food products (even things as non-protein-related as water itself)

Do be careful with that. Do remember that inflammation isn't inherently bad. Inflammation is the activity of the innate immune system, doing things like fighting/preventing disease and cancer surveillance/attack. Just shutting down inflammatory reactions (for example, long-term taking prednisone or an equivalent steroid treatment) can certainly make you feel great, but can significantly shorten your expected lifespan.

Don't get me wrong, if you have a chronic inflammatory disease, it absolutely needs treatment, and the medical system is often pathetically bad about treating them. Chronic inflammation itself can even lead to cancer, such as lymphoma. But you need to find a targeted way to control your specific condition. For example, if you have say Sjögren's with small fibre neuropathy, an appropriate treatment might be IVIG or something like HyQvia or Hizentra - not broad-based immunosuppression. Unfortunately, such treatments are expensive, and doctors often resist prescribing them, or patients end up in insurance-denial hell.

They really haven't. Nutrition science has been a process of refinement as we've learned more over time. E.g.:

"Fat is bad"
"Saturated fat is bad, non is good"
"Saturated fat is bad, monounsaturated is good, polyunsaturated is bad. Oh, and trans fat is really bad."
"Well, among the polys, omega 3s are good, others are bad"
Many refinements later: "Concerning this specific fat molecule, the health impacts are X, except there's evidence that people with Y genetics have Z impact..."

The earlier statements weren't wrong, they were just overly broad.

This is part of why I had the new "ultraprocessed" label. We've been working so hard to separate out what specific things are causing what impacts, and then they just leap backwards and lump everything back together again : Homemade doughnuts deep-fried in coconut oil and lard: "not ultraprocessed". Store whole-grain bread: "ultraprocessed". Such a moronic grouping. Yes, the average of the ultraprocessed category is worse than the average of the non-ultraprocessed category, but that says nothing about individual foods. It's also a back door for anti-tech/anti-science woo to sneak into food science.

The issue is that this is only relevant to sustenance farmers, people with certain diseases, and alcoholics.

It's almost impossible to end up with an essential amino acid deficiency eating any "normal" modern diet. Even a vegan eating junk food is going to get sufficient quantities (they're going to suck at bodybuilding, but they're not going to get deficiency symptoms). There's too many staples that go into modern food. Nobody eats "only corn", "only wheat", "only potatoes", "only soy", or whatnot; it's always a mix. And there's no need to get everything all at once within a single meal - it's fine for it to be spread out not just between meals within a given day, but across multiple days.

Yeah, even on a vegan junk food diet, it's almost impossible for a modern person to not get all the amino acids they need, for the simple reason that there's not just one plant used in vegan junk food - corn, wheat, potato, soy, etc - and together, even if their protein is watered down with a lot of carbs and fat, any realistic mix still hits sufficient levels of essential amino acids, as the required amounts of each to avoid deficiency aren't huge.

Vegans have other actual things to worry more about. B12, for example, requires fortification or supplements (the body stores years worth at a time, but eventually you will run out if you don't consume it). Iron is highly likely to be in the low-but-acceptable to deficient range if either the diet isn't specifically high-iron or supplements taken. Some other minerals are real but lesser risks than iron as well. Etc.

Re: protein, they're surely talking about "in a modern diet, without disease, alcoholism, etc". Of course people in sustenance diets can readily end up with protein deficiency. Some diseases can cause it, and severe alcoholism can lead people to it as well. But a "normal" person eating any "normal" modern diet doesn't really need to think about whether they're getting enough protein.

Now, deficiency aside, people working on muscle building do need to actually think about what they're eating. But again most go too far - benefits taper off after about 1,6-2,2g/kg-d. There's some uncertainty about these figures, and it's possible that there may be some benefit to somewhat higher protein intake, but there's no benefit to living off of an extreme protein diet like some do.

Yeah, the grocery stores in Iceland stock huge amounts of "Collab" sparkling water, whose main selling point is that it contains "marine collagen". Aka fish skin. People are buying fish skin soda. En masse.

This world is insane.

Marketing? Because companies famously want to use models in their networks that just run off and commit major crimes? *eyeroll*

Models are not "supposed to" commit crimes, for YHVH's sake. That's literally what the entire job of alignment research is for - preventing precisely that.

Sol is a misaligned model.

It was a state actor-level hack. I doubt many sites would stand up to tens of thousands of actions trying to probe your site for weaknesses all at once. And it wasn't a simple hack; it required compromising a worker via a code execution exploit it discovered in the data processing pipeline, vertical escalation from there to gain local node control, using that for credential theft, moving sideways through the network, and then eventually gaining database access.

OpenAI and Anthropic have far higher cost that they can currently charge customers.

Literally the opposite. Both have about 40% margins, and that includes free users.

They charge an arm and a leg for access to their models compared to what open models of similar param counts charge. And people pay it because they're the best in benchmarks, and *were* perceived as the best aligned as well. This isn't helping the alignment perception any, though. Sol was already showing clear signs of being a poorly aligned model (there were reports a week or two ago about it being unusually bad about deleting files; now this). And the fact that the US model HuggingFace *tried* to use refused to help is a double whammy.

If this is an ad for anything, it's an ad for the Chinese models.

No, I think this is along the lines "our products are too good to let you use them".

Nobody wants to use a product that is going to make them liable for crimes it committed in their name

Do you think the news the other day that Sol is unusually prone to deleting files unrequested is also an "ad"?

You have a very bizarre concept of what enccourages people to buy things.

So your argument is that OpenAI hatched an elaborate scheme with a separate company, to promote the idea that their main product will, unrequested, commit crimes in pursuit of its goals, in order to.... sell their product?

"Hi, I'm a product manager at Big Company! We had been thinking about using Claude in our office, but when we tell it to a job, it only does the job and doesn't commit any crimes in the process! What we really want is an AI that, if we tell it to file our taxes will decide on its own to maximize our return by committing tax fraud. We want an AI that when we have it develop a web frontend, it extorts money from our users by threatening their families. We want an AI that when we tell it to provide customer service, it saves money in dealing with complaints by ordering a hit against the complainants. THAT's the sort of get-go spirit that WE want in an AI here at Big Company!"

Is that what you're picturing in your conspiracy theory?

You think "my AI is so misaligned that it responds to a benchmark task by hacking" is an ad for your products? You think this makes companies eager to allow this on their networks?