Model interpretability work has advanced a lot. Arguably we already can explain AI decision-making better than human brains.
752 karma · joined May 16, 2025
Model interpretability work has advanced a lot. Arguably we already can explain AI decision-making better than human brains.
1) If the AI companies do end up running half the economy, we will have discussion for the rest of the human history about how the public got scammed by not being able buy in earlier at a lower price and how the late IPOs set in stone the oligarchy.
2) If the AI companies crash and burn, we will have discussion for several years about how everyone involved in running and financing them is a scoundrel who needs to go to jail for scamming us by selling us stock.
You believe that the owners of the leading AI companies stand to become owners of most wealth. Furthermore, that we are at an inflection point where the value of these companies rises so rapidly that delays in index investment will set in stone a permanent inequality, where early tech VC and other private funds own a huge portion of the economy. The few-$T downside risk of AI bubble popping this year feels to you like a minor concern compared to index funds being shut out of this wealth due to some arbitrary rules, which have been changed before and can be changed again. Delaying investment in these huge public companies feels like a more dangerous decision than buying in when they become public.
In short, there are two possible stories here:
1) Wall Street is AGI-pilled and thinks AI companies will be worth many trillions of dollars
2) Wall Street expects the AI bubble to pop and is trying to make the public into bag holders by selling a few hundreds of billions of dollars in the IPO
I think the second story actually doesn’t hold together, because Wall Street is making a bunch of correlated bets. The IPO cash is just one more source of capital, and much of it going to be used to make investments which are also correlated bets.
Twitter has about $13B of debt, and about $1.5B of annual interest payments (that’s how much cash it actually needs to come up with this year). SpaceX has a planned IPO market cap of $2T and plans to raise $75B cash during the IPO.
Nevertheless it is part of a pattern of weird deals in Elon’s companies. He’ll do anything to move the goalposts and turn his failures into successes. There is no norm he won’t violate, no boundary he won’t cross.
I agree Anthropic faces some risk they could get commoditized, but on the other hand if things go well they could end up leading adoption into more industries. There are upside and downside scenarios. Recursive self-improvement is obviously an important unknown and could lead to winner-take-all.
Theoretically, if training more expensive models stops resulting in better capabilities or isn’t economically viable, the labs can shift gears into making profit on old models. A lot of future growth is priced in so this would lead to a collapse in share price if it happens anytime soon.
There’s a story out that Anthropic might be profitable this quarter. This is in one sense bad news - it means that the company wasn’t aggressive enough about acquiring capacity last year, because they didn’t foresee how fast their inference business would grow. Anthropic is now forced to make suboptimal choices about serving existing users vs. training the next model (need to scrounge for capacity by paying other players like SpaceX). And as a Claude Code user I feel like I’ve been affected by that, what with the random outages and performance degradations.
The industry is actually doing real work on water issues in response to these complaints. Big tech companies are funding projects that will allow them to replenish more water than their datacenters consume. This isn’t actually that hard of a goal for them to meet, because as we know, the amount of water we’re talking about isn’t much on a national scale. Regardless, this will mean companies making some positive change in the communities where they build datacenters.
Anyway, all of this is a distraction compared to the real problem of carbon emissions. It confuses me that environmentalists are getting sidetracked by the water use distraction here when more natural gas and coal plants are coming online.
However, I think it was on the phone with the company waiting on hold for about four hours total to process the claim. That was without any dispute or confusion, it’s just how long the process took. The person on the other end of the line would ask me one question, then put me on hold for another 15 minutes before asking the next question. I got the impression they are actually handling multiple customers at once and switching between us. I can only imagine this system is designed to deter folks from making claims. I was afraid to delay the call any longer by inquiring about whether there were multiple replacement models I could pick between. I received a $600 model from a manufacturer I wouldn’t have chosen.
If I’d paid $600 for that home warranty, I would not be a satisfied customer. I certainly wouldn’t recommend home warranties as a time-saver. It would be much easier to directly purchase products and services from people with some incentive to obtain my business.
Counterargument: just because the problem can be fixed without training, doesn’t mean training isn’t a possible solution.
Arguably, there’s a disconnect where the people who lead civic organizations don’t have a great deal in common with the median member. They might be wealthier and generally more plugged in to power structures. They might not support policies that are in the best interest of members they represent, especially people who have a hard time representing themselves.
Anyway, if your goal is to get a policy enacted, it’s not enough for your policy to be theoretically good for the median voter. You need a winning political coalition.
1) same business logic implemented in two different places, with extra code to sync between them
2) fixing apparently simple bugs results in lots of new code being written
It’s a sign I need to at least temporarily dedicate more effort to overseeing work in that area.
I somewhat agree with the AI psychosis framing of the OP. It takes some taste and discipline to avoid letting things dissolve into complete slop.
If you have a lot of trash, most economical option is often to go to the public transfer stations or landfills in the Reno area (but that only works if they are open).
Also, there are services on the side of the highway that accept trash for $N per bag. Only give your trash to someone if you can see the dumpster it’s going into and the dumpster is not full. There have been scams where people charged to accept trash and then just left it there to get blown around the desert. Alternatively, you could drive your trash all the way home and let your local utilities handle it. But when I’ve had my cargo trailer piled with leaking garbage bags I’ve wanted to get rid of it ASAP.
I thought the prime bet was that the winning lab who reaches takeoff through recursive self improvement will make a galactic superintelligence. Not saying I believe this but the people running the labs do. Under this scenario if you are a few months behind at the pivotal time you might as well not exist at all.
If you want to run local models, another advantage is Apple’s unified memory architecture. The biggest Mac mini has 64gb ram and Mac Studio has up to 512gb. Compare this little box to what monster Nvidia gpu system you would have to buy to get the same memory there. And how much your PG&E bill would go up. That doesn’t account for the shortage of basic $600 Mac minis though.
> Parents are solely responsible for bringing their children on stolen land. There were indigenous children living there that were murdered.