shimon

Yes. Altman was just pointing out something about economies of scale.

You don't have to agree with his values (e.g. you could believe that individual ownership of computation is more important than overall efficiency) but the statement here is logical and more or less obvious.

Can confirm. Our household is involved in raising two natural intelligences. They are quite expensive to maintain. Chances they will offer positive economic return are unclear. The value proposition on comptuer-based intelligence, by-token or through hardware ownership, seems much clearer.

I have to admit I am a sort of enthusiast for the natural intelligences, though.

Did you get any interpretation of the imaging? It sounds cool (I'd take an ultrasound selfie) but it has to be medically useful for something before enough doctors grade enough images to kickstart the data collection for diagnosing more.

LLM/AI growth is the major driver of usage growth on all the clouds.

Actually no, LLMs (preferably tool-using LLMs that can themselves do web searches and price lookups) are a great way to shop, especially if you're looking in a category and don't know exactly what item you want.

You just say "I want to find some headphones" and it makes you some recommendations. Or it helps you nail down what you're looking for first and then gives you options at various price points. I've found this useful when shopping for cars, computers, tourist activities, and much more.

At what point does the use of clues to uncover the identity of a criminal cross the line from solid detective work to doxing? /s

Can you explain the joke in that username?

I totally agree, but it is reflective of a shift in how cultural products are designed and selected that is in my view the more interesting story here.

When selection is personal, you have a smaller set of choices and will respond to how things feel. If you're walking down a street looking at 3 restaurants, a surprising element in one might speak to you in a way you can't even explain. On the other hand, if you're just browsing in maps, you'll be going by ratings that reward a more generic notion of quality. That type of selection is algorithmic rather than personal.

More parts of our lives have been algorithmized than even before, and we generally like this because selecting things is hard. But algorithmization can drive us toward local maxima, where the pleasantness of familiarity draws us away from selecting or creating something truly personal. Of course, it can also draw people into arcane and corrupt niches like conspiracy theories.

The algorithmizations of physical spaces and social connections are probably the newest and most distressing. These realms used to be almost entirely localized; the idea you'd pick your friends or spouse by algorithm used to be rare and is now dominant. But personal relationships are incredibly nuanced and the algorithms are inadequate to the diversity of human desires. Yet they're shaping our desires in a way that is preventing many people from finding their happiness.

I'm particularly impressed by your choice of "up there with Euchre" as a metaphor to explain Crokinole. It's like you wanted to make it relatable for people in a larger geographic region, but only a little larger.

In many cases, a simple off-the-shelf smart plug would suffice.

On what factual basis can you claim that adverse events are primarily driven by genetics?

On the face of it this seems ludicrous. A baby born to a mother living in a high-risk environment but then adopted by a low-risk family would likely do far better in their life than the inverse.

Yes they also typically buy reinsurance (insurance that covers individuals whose claims exceed some amount in a given year, like say $150,000). The company will have some people in HR who administer this program but they aren't duplicating a lot of what the insurance companies do -- that's why they pay for administrative services.

They rarely have actuaries on payroll but there are several boutique actuarial consultant who serve these companies (e.g. "given the geographical distribution of your employees and the higher use of services X, Y, Z, you would save $xxx if you choose insurer A over insurer B").

Almost every sufficiently large employer does this. Instead of paying premiums to an insurance company which owns the risk (and also makes a profit from that) the large employer pays the actual costs of care, and a percentage to the insurance company for the administrative work (using their provider network, claims processing systems, etc.).

Insurers don't want to pay for unnecessary treatments, including equipment. Maybe your provider could have done a better job selecting the right mask or settings for the CPAP? Maybe they could have trained you in their office?

I'm not saying the surveillance and payment-held-hostage model is the best, but it does at least attempt to provide some useful incentives.