learning how to be productive with LLMs is a process of learning just how audacious or unreasonable you can be when talking to them.

Some very cool work from my colleagues at Goodfire! It's amazing to see how models build their internal representations of the data they're trained on, especially when you have data from distinct domains like gene sequences.

Join millions who have switched to Grok.

anyway, that is all. go forth and enjoy quality food, and stop ordering daily private taxis for your seed-oil fries.

occasionally using some cold nut/seed oils for a dressing is also not terrible; usually one just adds a bit for flavor, and they can be nice. (you can also store them in the fridge).

you don't have to be neurotic about it (I'll have some fries now and then) but don't be seed-oil-maxxing, and keep the rough numbers in mind. there is no reason to use them at home; avocado oil is relatively neutral and this (or rarely, canola) can be used for Asian dishes.

all of this is to reiterate: the details of how bad PUFAs are in isolation *don't matter* to the question of whether you should consume seed oils. if you care a lot about your health, you should consume them only sparingly/occasionally.

why is EVOO better? just do a search in Google Scholar or your favorite Deep Research product. but it's very high in polyphenols and linked to many positive health outcomes. I use it daily, personally.

this brings us to issue 4: opportunity cost. if you are using seed oil to cook something, you are not using extra virgin olive oil or avocado oil. is it better than butter? well, idk--see above. but NO seed oil is better than EVOO, especially tested/verified brands.

you could have a bit more canola oil (7.7 TBSP) because it's lower in PUFAs. and again, note that this is with unusually high consumption of fish that is unusually high in omega-3s.

hardly any, as it happens! you could have 2-3 TBSP of safflower, sunflower, grapeseed, corn, cottonseed, soybean, or rice brain oil a week, assuming you have no other sources of omega-6 in your diet (unlikely). more than this, and you will exceed the recommended 4:1 ratio of 3:6.

if you use seed oils for cooking a lot, it is almost impossible to maintain omega-3 balance. Westerners already eat very little fish, but let's assume you eat 16 oz of wild salmon per week (I know you don't, but maybe you should). how much seed oil would result in imbalance?.

does this happen if you just use canola oil once in a while for home sauteeing? not so much. the biggest risk is restaurant/fast food, where oil is re-used. BUT! using these oils extensively at home leads to issue 3:.

also are incorporated into cell membranes, altering their function, damaging immune function, increasing heart disease risk, decreasing hormone response (including insulin), etc.

usually, these oils are used for deep frying or industrial snack production. at restaurants, the same oil is used multiple times for deep frying, which of course dramatically increases the oxidation. the resultant aldehydes are mutagenic and increase cancer risk, and.

first, unless you buy the cold-pressed, opaque-bottled, well-tested versions of these oils (which is sometimes available), they're usually bleached, deodorized, and stored in clear plastic bottles. we know this means that the PUFAs get oxidized, but the real issue is with #2:.

what matters is the reality of how you personally are going to encounter seed oils. there are four problems:.1. processing/storage/oxidation.2. culinary use.3. omega-3 balance.4. opportunity cost.

the literature around nut and seed consumption seems to show that these foods are consistently great for health, despite their PUFAs. so maybe these PUFAs are not inherently bad. but again, that doesn't matter much. so what does matter?.

first, are there good arguments about seed oils *by themselves* being inherently bad, regardless of processing, GMOs, etc.--that is, due solely to their PUFA content?. honestly, maybe, but I am not super confident in them. but my point here is that actually that *doesn't matter*.

it's also a great test of how good you are at integrating contextual information from multiple domains/realistic assumptions about real life, and doing first-principles reasoning with it. here's what I mean:.

the whole seed oil thing strikes me as a great test of how good your reasoning/need for cognition is. are you just going to take every trending argument at face value?. it's so much simpler (and at the same time, slightly more complex) than everyone here makes it out to be.