The administration's planning is much more along the lines of, Will this look cool when they announce it on Fox News tomorrow? If you think there's much beyond that, you're ascribing strategic clarity where there isn't any. They're continue to flail around and TACO until they have a result they can present to MAGA loyalists as a success, regardless of actual merits.

It's not a question of ethics. It's a question of competence.

This was my desktop computer responding to questions in English, not some fancy server in a massive Google data center. Who cares if what it says isn't reliable? Being able to converse with my CPU in English is like having a conversation with a dog!

These are the data center issues as I understand them, in ascending order of importance:

  * Water use: Almost always a red herring or non-issue, unless the DC is being built in an area with water shortages. DC's use a lot of water, but their use is negligible compared to many other industries.

  * Neighborhood appearance: They're not particularly pretty to have in your back yard, but much less ugly than, say, a factory. They're not inherently polluting.

  * Power draw: This is a legitimate concern as DC's use an enormous amount of electricity. In the short run, it could make sense for deep-pocketed investors to subsidize residential or non-DC power consumption to keep everyone's electric bills from skyrocketing. Longer term, power companies will need to build much more generating infrastructure. I'd love to see a carbon tax to encourage the construction of renewable (or nuclear) power. Sadly, the current US administration seems intent on vice-maxing and ruining as much as they can for future generations.

  * AI-driven job displacement: I think this is the real worry people have. The water use thing is an excuse people are looking for to oppose AI.

My optimistic take is that AI companies won't in fact capture all of the value from automation, because they'll be competing against each other, and against open weights models. But who knows? Maybe a single company will achieve Super-AGI first and they'll own the world. I doubt that will happen, but this is what they're aiming for, and a lot of the money invested only makes sense in light of that goal.

And even in my optimistic scenario, the job disruption will be quite real. New jobs will be created as other jobs are lost to automation. That's well and good after things have settled, but it is very disruptive to people's careers and ambitions in the mean time.

LLMs process language. I'd even go so far as to say that LLMs "think" and "understand", or at least, they produce a facsimile of thinking and understanding such that it's useful for us to reason about LLMs as if they think and understand. We're not used to interacting with a non-human entity with the capability to process language, so it's easy to ascribe human traits to these things. But their "minds" (insofar as they have anything like a mind) are completely different from ours. These things have language without consciousness.

Chimpanzees are conscious. Dogs are conscious. Maybe ravens and cephalopods? Who knows. These animals do have minds much like ours. Higher order animals are conscious even if they don't have language.

Edit: Jinx! Gracana beat my reply to an 11 hour old comment by four minutes.

This is an administration that has neither of those.

Ideally, driverless cars will one day be better drivers than humans and this will save tens of thousands of traffic deaths per year. Holding up progress because cars will be confused in extremely rare or improbable situations will cost more lives than it saves.

Imagine a busy intersection where all the cars fly past one another at 40 miles an hour without stopping but none of them crash. Humans can't do this, but machines could, if, and when the technology gets there. To be clear, there's still a way to go.

In that light, it's probably a good thing.

It just doesn't have much range: only about 25 miles on my 2018 model. Newer models advertise up to 44 miles on EV.

This has been a perfect car for my use case, but the big caveat is my short commute. If your daily commute fits inside that short range (or one way commute if there's a charger at your workplace), this can be a great fit. A+++, highly recommended.

If your work commute is significantly longer than a PHEV's battery range, or if you don't have a convenient place to charge it, then it's a much less attractive proposition.

Because of course it is.

Very "intelligent" people can use those smarts to justify or rationalize all kinds of crazy stupid decisions.

Having said that, you're right to be suspicious of commercial services, even that you pay for. Someone can found a startup with a strong commitment to customer privacy and the best of intentions, but a few acquisitions or near bankruptcies later, those commitments will go out the window.

That is a tough sell in the current environment. It's a regressive tax, so opposed on both ends of the political spectrum. People on the far right don't believe in climate change, and people on the far left don't believe in market efficiency. With 20% of the world's oil flow crimped in the Strait of Hormuz for who knows how long, higher energy prices is the last thing people want to contemplate.

In the longer run, a carbon tax is the best option. The fossil fuel price shock is a strong signal to produce energy through other means. There are major engineering initiatives around developing cheaper and safer nuclear energy. and it's cheaper now to deploy a solar farm than a coal plant.

A carbon tax would raise money to pay off national debts and encourage consumers and producers to figure out the most efficient way to accomplish their needs while minimizing their carbon footprint. It's a tough sell today, but this is they way to go for a better quality of life tomorrow.

What should Anthropic do in this case?

Anthropic could immediately make these models widely available. The vast majority of their users just want develop non-malicious software. But some non-zero portion of users will absolutely use these models to find exploits and develop ransomware and so on. Making the models widely available forces everyone developing software (eg, whatever browser and OS you're using to read HN right now) into a race where they have to find and fix all their bugs before malicious actors do.

Or Anthropic could slow roll their models. Gatekeep Mythos to select users like the Linux Foundation and so on, and nerf Opus so it does a bunch of checks to make it slightly more difficult to have it automatically generate exploits. Obviously, they can't entirely stop people from finding bugs, but they can introduce some speedbumps to dissuade marginal hackers. Theoretically, this gives maintainers some breathing space to fix outstanding bugs before the floodgates open.

In the longer run, Anthropic won't be able to hold back these capabilities because other companies will develop and release models that are more powerful than Opus and Mythos. This is just about buying time for maintainers.

I don't know that the slow release model is the right thing to do. It might be better if the world suffers through some short term pain of hacking and ransomware while everyone adjusts to the new capabilities. But I wouldn't take that approach for granted, and if I were in Anthropic's position I'd be very careful about about opening the floodgate.

https://charliesykes.substack.com/p/a-vivid-snapshot-of-trum...

Some highlights include a $580 million dollar bet on oil futures 15 minutes before Trump made the announcement of talks with Iran, which the Iranian government denied actually happened.

Naturally, political appointments at the SEC are preventing investigation.

That range is a significant caveat. If your round trip commute (or one way commute, if you can charge at work) is outside the electric range, then you'll be relying on gas every day. In my situation it's worked out extremely well. I charge at home and only need to fill the gas tank about three or four times a year.