ML Engineer at KoBold Metals. Former founding engineer @ sematic.dev . Ex Cruise, passionate about making ML Infra accessible. Opinions are my own and do not necessarily reflect those of my employer.
I only have one car at a time (not married). I frequently visit family that lives about 80 miles away that live in a charger desert. I used to own a Chevy Bolt and it could get me to those family and back on one 100% charge with a little bit for grocery runs etc. while there. Even that I wish had like maybe another 20 miles of range lol.
The only reason gerrymandering works as a tactic in the first place is that neighborhoods DON’T have a statistically uniform distribution of the population. So “Steven and Sally live in the same neighborhood” is not a very good way to look at issues like these.
Switching cost for a DB and switching cost for an inference API are very different things. They’re going to have to create lock-in with something other than model intelligence if they want to play Oracle’s games.
Can attest that Tidal is basically completely satisfying as a Spotify replacement for me. I use a dedicated podcast app and buy any audiobooks I want from Libro.fm Haven’t looked back.
Still continue to slowly build my CD + mp3 collection in case this wonderful era of “cheap subscription for almost all the world’s music ad-free” doesn’t last forever.
Why would I hire Alice to work for 4 days a week when Bob is willing to work for 5 at the same price (assuming both are equally productive, even if AI has made that productivity extremely high)? So long as there is an oversupply of qualified workers, I doubt working hours will change. To do so would require social contract changes like for the standard 40-hour work week.
A much closer analogy than the pen one is a SAAS company having an EULA. As annoying as EULAs are I don’t think it counts as being malicious or subversive if you are very open and clear with your prospective customers what that EULA contains?
Seconded. I truly consider that podcast a gem of our culture. I know of no comparable resource for generally curious people willing to try and think deeply about things.
The “recursive” part comes from the fact that you have an AI which was developed by an AI (that was developed by an AI (that was developed by an AI (…)))
Less bad, yes. But still bad. There will definitely be demand for the compute, the question is whether it will be enough to keep the value of it at the levels you’d expect if the tenants were in a monopoly/duopoly world. Competition puts downward pressure on margins and that’d translate into pressure on data center leases.
100%. I have sometimes wondered if China is playing the long game with the open weight models trying to tank the margins of the US labs so these profits don’t materialize and the US economy (currently predicated on the net that they will) suffers. I think this is a coherent strategy beyond just “don’t let the US control AI as a strategic asset.”
I think the point is probably to help convince people that the watermarking doesn’t perceptibly impact quality, which is a concern some people have (whether well founded or not).
Yeah, I hated all those Quora users that would just spew out semantically meaningless slop like increasing an important bound for the Riemann hypothesis.
They need him to continue evangelizing how to get a single individual to burn $87k/month so the corporations who have to actually pay usage pricing will actually make them bank :)
I don’t know that I’d call it that much of an outlier position. Compatibilists, functionalists, and plenty of physicalists in general would deny that qualia exist or that they’re a well-defined notion.
If you buy the argument “If A can do X but B can’t, then B is less intelligent than A”, then you have just as strong of an argument that LLMs are more intelligent than cockroaches. When’s the last time you saw a cockroach solve an Erdös problem?
If I ask my kid (early elementary) what he learned last school year, I can watch him apparently genuinely trying to think of some examples and not being very successful. If I ask him something I know he learned he’ll gladly say “oh yeah!” and tell me about it.
In terms of whether they’re overvalued: probably. But any valuation should also take into account the value they have to x.ai (also under SpaceX) as a source of training data for coding models.
This piece made me kind of fall in love with mechanical watches. The Orient Bambino I’m currently rocking on my wrist probably wouldn’t be there without it!
You are describing where I want to be when my kid graduates (years from now), mandolin and all! Low cost of living, contributing to the community, enough time for passions… Not able to get there yet but it’s what I’m working towards. I just hope LLMs don’t eat my current career path before I can achieve it.
Chess is a game where we play and observe for the sheer joy of mastery. Mathematics has some of that “done for its own sake” aspect but not as cleanly. There’s always the chance that some applied mathematics might help something in science or engineering. Whether we care more about process vs outcome is now more important than ever before.
I’m not saying they should stop building things (I doubt any argument in that direction would land anyway). But more that they should follow through on the whole picture of the future they’re working on. Anthropic’s brand leans a lot more on trying to be the good guys than other big labs, which is why I think they’re worth addressing in particular.
Anthropic employees: I believe many of you actually do think your work is for the betterment of humanity. That’s great! If you truly do believe that, it’s time to start demanding your leadership lobby for real, structural solutions to unemployment that leave everyday people with a say in how companies like yours operate. “The market will find new jobs” is not as strong of a historical pattern as “the concentration of power harms the people without it.” It’s true that AI might not in fact automate the majority of labor in the near term. But if it does (and working at Anthropic implies you likely think it could), then work towards a version of that future that’s actually palatable. It would be easy to let your comp upside soften your objective evaluations. Don’t. Think you’re working for a better future? Put your money where your mouth is, and ask your leaders to do the same.
Sadly this won’t happen for the same reasons that the median salary hasn’t increased with increases in productivity over the past few decades. You are compensated (or given time off) based on what your employer can get away with, not based on what they get in return. The system is designed to accrue all excess value to shareholders, not employees.