Some quick-mafs.

Assuming the use of a high milk yield breed like Holstein Friesian cows.

18KL per lactation: https://www.australiaslivestockexporters.com/holstein-fresians-dairy-cattle/

About 217 ml of milk, 23.6g of cheese (equiv to 236ml milk) and 21.4g of yoghurt (about 21.4ml of milk) are consumed per Australian per day: https://www.statista.com/statistics/1143391/australia-dairy-mean-daily-grams-per-capita-by-food-subgroup/ = (approx) 474ml = 0.47L

Australians live about 83 years, approx 30k days.

30kDays * 0.47L/Day = 14100L of milk over a lifetime. Or a bit less than one calf.

The average Holstein cattle has parity of < 2.7 (https://pmc.ncbi.nlm.nih.gov/articles/PMC8369829/). So we can estimate something like one third of a dairy cow per person, and a little less than one calf.

The primary effect of diary on cows life wouldn't be the cow that's milked, it'd be on the calves. I suspect it'd be more than 1/2 of a calf per human milk lifetime.

Even in markets where a positive resolution doesn't literally result in the death of most people alive, this can be a concern.

I've been tempted in the past to place money on nuclear war markets, but the fact that most prediction markets are pegged to USD makes me hesitant.

As an Australian, if a nuclear war breaks out, I would predict a much higher likelihood for my local government to keep meaningfully functioning than for nearly any other government in the world. Therefore, the value of Yes on a nuclear war market for me is meaningfully higher than that of most of the rest of the userbase.

But since the market resolution is pegged to USD, I can't meaningfully express that. If I could get some financial product that converted USD expressed smart contracts to AUD expressed ones (maybe backed by FX futures?) that would help, but for obvious reasons nobody is supplying these.

I'm not really seeing the point of AI augmented human labour here.

It seems like it's meant to fill the gap between now and the production of either generalised or specialised macrorobotics, but it seems to me that that niche is better filled by existing machinery.

Why go through the clunky process of instructing a human how to do a task, when you can commandeer an old factory, and repurpose some old drones to do most of the work for you? Human beings might *in theory* have a much higher ceiling for precise work, but realistically you can't micromanage someone into being good at a physical task; they need to build muscle memory, and that's gonna be hard to come by with the constantly changing industrial processes a super intelligence would presumably be implementing.

On the other hand, you could macgyver old commercial machinery into any shape you want, quickly spin up a virtual training environment, and have an agent trained up on any industrial process you want in presumably minutes.

I think you might be assuming that industrial robots are hard, just because humans are bad at designing them. But I reckon a little bit of superintelligence would go a long way in hacking together workable robotics.

"No matter how intelligent you are, you cannot override fundamental laws of physics."

Call me crazy, but I don't think this is necessarily true. While it's obviously true, by definition, that you can't break the fundamental laws of physics, the laws of physics almost certainly don't neatly map to the mathematical symbol games a bunch of us monkeys derived to describe them. It would be wholely unshocking to me if, for example, something we view as "fundamental" like the laws of thermodynamics are more an artifact of our models than a true facet of reality.

I definitely agree with the rest of your point though. So long as humans aren't on the pareto optimal frontier of the energy/intelligence tradeoff (and I really doubt that's the case), you should expect intelligent machines to be strictly superior in capabilities to us in all meaningful ways.

Very impressive.

I tend to avoid using LLM assistance, since nowadays I mostly program in a hobby capacity, but I'm especially impressed by how well it navigated a pretty obscure programming language (Odin) in what's presumably a fairly messy codebase (No judgement towards you! I've just never seen a videogame with clean code, I think it just comes with the territory).

Based on what I'm seeing though, it seems fairly squarely cemented as a tool for people who already understand how to program for the near term future.

There's a strange lack of discussion about grandparents in this thread.

I would love to see complementary data on the time that grandparents spend with their children. While parenting norms have definitely become more intensive, I can personally attest that things like multigenerational housing make that significantly more sustainable.

It would be much stronger evidence for parenting intensity norms changing if the sum of Parent Time + Grandparent Time + Tertiary Caregiver Time increased.

Side note: Lest Australians think we have it much better: https://mypolice.qld.gov.au/farnorth/2022/09/06/child-protection-week-4-10-september-what-age-is-a-good-age-to-leave-my-child-alone-at-home/

Is there a name for the general principle that doing boring things is more effective than doing interesting ones? It seems generally true in a lot of situations.

It seems to me that the sheer dominance of transformer models compute efficiency improvements is a horrible sign.

A large number of iterative improvements should be somewhat easy to monitor, a small probability of a sudden massive leap in capabilities is a lot harder. Especially since, a priori, I see no reason for all future compute dependant advances to happen at close to cutting edge levels of compute. 

It seems entirely possible to me that something like a future version of alphatensor could stumble upon one of these hidden gems of an optimisation, verify it with a level of compute it has unmonitored access to, and quickly dominate the competition in compute overnight.

Black swan events are scary.