mr_bailey
17d ago
Then don't use it, that's the whole point of a free market. Same for the DoW, I have no problem with them not using Anthropic (and from everything I've seen Anthropic was perfectly willing to let DoW unilaterally end the contract early and offered support for a transition period to a new vendor), the retaliation is just excessive, capricious, and completely outside the intent of the law in use (I suppose par for the course for Trump 2.0 admin).
Arguably, though, Anthropic's ethical stance is partly why they are the best, so I'm not sure this is so easily separable. Certainly it has been key to their recruiting and how they have been able to attract the talent they have and they are by far in the lead in model interpretability thanks to this focus. They have also argued in various papers that this approach has led to better model performance (most obviously in false refusals, which will be a thing until we decide we're ok with commercial models teaching people how to make chemical weapons, etc.).
mr_bailey
17d ago
Oh I'm sure that's what they are going for. Though the lesson could as easily be don't even start doing business with the US government which may not be the win they imagine.
mr_bailey
17d ago
This seems to be an entirely different claim though. Is the problem that Anthropic is insisting on certain contract terms around selling it's current products remain in place or that it won't generate a more morally deferential AI? The later seems to be what you object to, but, in theory at least, is not the crux of the current kerfuffle. Developing an AI within a consistent ethical framework is kind of Antrhopic's whole thing and arguably has probably helped them (certainly at minimum, with recruiting). Idk, mileage might vary, but I've found Claude to be pretty nuanced in it's opinions on the use of violence in the context of self-defense and police shootings at least compared to ChatGPT and Gemini which seem to be a lot more proscriptive and it is certainly empathetic to the users position. I'm not at all convinced on your claim. If you're looking at models likely to tell someone off, Grok or some of the chinese models are much more likely. GML 4.7 is too far off the frontier (or alt. to narrowly focused) for me to consider it a strong comparison point. If that's what you want/need/suffices by all means use it, but it's not a replacement (or if it is not sure why DoW is so focused on Anthropic).
mr_bailey
17d ago
A highly relevant aspect is that the government paid Lockheed to develop the F35 under specific contract. It's not exactly commensurate, but would it be a supply chain risk if SpaceX said it was unwilling to launch nukes into space?
mr_bailey
17d ago
That's an inherent risk, they could turn off the tap (to the extent they are able to, I don't believe Anthropic is actually running the hosting) whether they agreed to the contract changes or not. Anthropic offered a 6 month transition period gratis for the DoW to transition to a new vendor so does seem to be operating in good faith.
mr_bailey
17d ago
I mean what will happen is they'll get it from someone else. Otherwise, what exactly are they going to take? This isn't some factory or a warehouse full of inventory. Maybe they can take the current model weights and have something that's obsolete in 6 months. What they want is a Claude 6 that will do whatever they ask it to and for that they need Anthropic and it's employees to cooperate and without North Korea levels of oppression there's only so many levers they have.
mr_bailey
17d ago
I mean, current kerfuffle aside (which you have to admit is highly contingent, there's no way anything like this plays out if Trump isn't president), Anthropic seems to be doing really well commercially? It has the fastest revenue growth of any of the AI companies (and on current trends would overtake OpenAI in the next year or so) and seems to be the leader in integration into workflows etc. Given it's rather paltry free tier adoption and rather high API rates it's likely already significantly profitable on marginal inference basis. I'm not at all convinced that it's ethical stance is hurting it (and it's virtue ethics approach may in fact relate to why it tends to have lower refusal rates then OpenAI and Gemini). I'd be curious on a poll of career soldiers on their opinions on autonomous killing robots (the point of distinction, Anthropic did not prohibit the AI from helping kill people, only doing so completely autonomously), I'd don't think they'd necessarily want to be out of a job.
mr_bailey
1mo ago
Maybe the actual product here is something like proton mail (I think that was the one). E.g. this AI doesn't keep logs so there's nothing to subpoena (tripwire disclaimer, or host in a friendly jurisdiction). Or we could go for a robot priest, Confessional AI, only the machine god can absolve you of your sins.
Does seem like there's some room for nuance here. Lawyers communicate via a variety of means, e-mail, etc. (though a lot do seem to use a specific messaging services), but I would expect assistive technologies would be protected (screen reader, interpreter, translator, what about that old phone service for deaf people, I believe that was not subpoenable) and to the extent one could do a service that helps you interpret the legalese from your lawyer how much does it differ from those?
mr_bailey
2mo ago
Yes, arguably the Inuit inhabitants are colonizers as well displacing the earlier inhabitants which come from a separate wave of settlement (and there was at least some small amount of friction with the Norse inhabitants before they died out too).
mr_bailey
3mo ago
Agreed on "I can't help but suspect the man funding 3 kids feels pretty different than the one funding 3 retirees." Especially since the later will mostly be funded by taxes (the whole problem is that lots of folks aren't having kids and the state has more or less taken over the inter-generational transfers to the elderly), whereas the former is more mixed (a lot of 'funding' there is paying for your own kids, though taxes do come into play for schooling etc.).
mr_bailey
3mo ago
You're misunderstanding the 1.6 number. That's the dependency at a given time not some notion of over a lifetime. 2.1 kids would, on it's own, give you a dependency ratio of <0.5 (2 parents, working for 40+ years, vs. 2.1 kids for 18 years each). If based purely on children a dependency ratio of 1.6 would ~ match up to an average of 7 kids per family.
For historical comparison, the US peaked at 0.68 in 1961 (baby boom) fell to it's lowest point 0.48 in 2007 and is currently 0.54 (baby boomers retiring). The highest in the world right now is 1.05 in the Central African Repubiic (lots of kids). 1.6 is really high...
mr_bailey
7mo ago
It increasingly feels to me like the Tyler Cowen's of the world are right. That the impact will be large, huge even, but take a lot more time to play out then the boosters predict. It will take time, not only for the tech to improve, but for people and companies to learn how to best use it and for complementary infrastructure and skills to build up. The parallels to the personal computer or internet seem increasingly on point, especially the dot com era. People were, rightly, astounded by those. And, for all the jeering pets.com got, and all the (mostly valid!) reasons it wouldn't work, it ended up mostly just ahead of its time. I and everyone I know buy my pet food through an online subscription or auto-recurring purchase. In 20 years I expect AI will be omnipresent in white collar work.
mr_bailey
8mo ago
I live near a large memory care facility, we get a lot of Silver Alerts from it. I'm ok with a text level of notification, but the actual alarm should be reserved for evacuation orders.
mr_bailey
8mo ago
There were probably just memorable and your brain converts being able to remember multiple storms as meaning they must have happened often. Most parts of CA really don't get major thunderstorms all that often. Once every 1-3 years sounds about right for where I'm at for ex.
mr_bailey
8mo ago
Yeah, a straight subsidy is better then whatever price controls CA keeps flirting with. There's a real risk that of breaking the property insurance market with those sorts of moves.
mr_bailey
8mo ago
His math is right:
"80 deaths 80 QALYS lost 365 2460 = 11 QALMS (Quality adjusted life minutes)"
80 deaths * 80 QALYS (generous, statistically prob. more like 60-70) lost * 365 * 24 * 60 / 330,000,000 => 10.19 (rounds up to 11 minutes)
Whether the population of the US is the right denominator is potentially debatable, but is not a priori crazy.
mr_bailey
8mo ago
We don't really have a good idea how different individuals will react, this guy seems to have pretty badly f*cked up his life https://www.theguardian.com/us-news/article/2024/aug/23/us-pilot-magic-mushrooms-plane-engines (tldr pilot takes some mushrooms for depression, 2 days later is riding in the jump seat of plane [as a passenger, not flying the plane] has a bit of mental breakdown and tries to turn off the plane's engines to wake up, the interactions with the crew are odd and suggest to me more of a break with reality [could be due to the shrooms, or not, hard to say definitely] then a murder/suicide but ymmv).
mr_bailey
8mo ago
It's another step removed from that, most of these studies are looking at Educational Attainment (e.g. highest degree received) which itself is a (highly) imperfect measure of IQ (which itself is an imperfect measure of General Intelligence 'g' which is the name given to the statistical observation that many different measures of what we consider intelligence correlate pretty tightly). The Genome Association studies are further largely using SNP databases which themselves more often only correlations to whatever loci are actually impacting things rather then directly impactful themselves.
mr_bailey
8mo ago
Things could get weird. The US (and North America as a whole) are net oil producers these days. Shutting off the gulf is such a large shock to the system you could see some pretty significant price divergence and shortages between regions as infrastructure limits could prevent fully arbitraging the difference. You could also see political impediments to price balancing as well (wouldn't put it past Trump to ban oil exports to keep US prices in check even if it's a huge blow to our allies) and most Canadian crude has to transit the US to reach the world markets. You could even see divergence in the US where West Coast is more exposed then East coast due to Jone's Act restrictions making it difficult to move oil around the Rockies.
mr_bailey
8mo ago
Mines, missiles, and drones? Civil ships are not exactly small targets and Iranian drones have seen some use in Ukraine. Could get ugly. It's likely the gulf states, East/South Asia and to a lesser degree Europe would be more affected then the US (who is, after all, a net oil exporter these days), so this hits US allies (and China, India etc.) much more then the US and Trump hasn't shown a high degree of concern about them...
mr_bailey
9mo ago
This interview with a Russian drone designer is from a few weeks ago, but really interesting from a perspective of how drone warfare has been evolving on the front lines.
mr_bailey
10mo ago
This is the popular conception of HOA's, but everyone I've been involved with has predominately existed to maintain shared infrastructure (in the single family home case often infrastructure that probably should be the city's responsibility, but the city is uninterested; they still get tax revenue but don't have to pay for it). It's a failure mode of HOA's but how common is it really?
mr_bailey
10mo ago
I'm president of the board of my local HOA. It seems typical for many in the area that it was put in place by the developer when the area was built out and that it exists at all is completely down to the city (that entirely surrounds the development) being unwilling to pay for anything. The dues of around $200/mo cover maintaining the roads/streetlights/signange, bridge over a 50ft ravine (it's hilly), landscaping along the roads and various maintenance activities with the 1/2 of the overall land that is 'dedicated open space' by demand of the city (the developer wanted to put in significantly more housing, that they were able to put in any was only after much negotiating from what I understand) which we are not allowed to say chop down dead trees in but also must keep from burning down all the houses and neighboring properties (a lot of weed whacking and brush clearing is involved). It would probably be dissolved if the city would take over the relevant maintenance (which it won't), it's all single family homes.
Our general challenge is getting enough volunteer board members (3), I don't think we've ever had a contested election in the entire history, and getting to quorum of votes (50%+1) in any election (last time was maybe 6 years ago?). The only somewhat contentious things we've had to deal with are parking disputes (no street parking allowed, we apparently can be forced pay to maintain the roads but the city gets final say on how they are used) so guest parking is limited. Technically, the by-laws give us the power to require (and enforce with inspections) that garages be cleared out enough to park two cars and the two cars are parked in the garage before parking in the driveways or guest parking, though we haven't yet reached the point of enforcing that. We have some other powers, but I suspect the only thing we'd enforce them around is planting large trees [eg. redwoods etc.], changing the exterior colors of houses, and hogging common areas [parking especially]. In general, the city does still get a quite significant amount of control over things, pre-me they tried to install a gate near the entrance of the community (it's like 15 acres or so, so plenty of room) and the city absolutely forbid it as in 'out-of-character' for the city and stated their policy is to allow only in the grandfathered places. So our power over this 'private' sub-city is limited.
The general idea behind the SAVE act seems not unreasonable and in some sense this is the time to do it (unlike in the even fairly recent past it's not clear given the current voting patterns that this particularly favors one party much over the other). That said the politics of it seem difficult to resolve.
More options
Context Copy link