TechTakes

This feels somehow tied to the whole "agentic" thing I've ranged about previously. Like, individual acts of violence are strictly destructive because the people doing it aren't sufficiently "agentic" to change things, even though American history is full of cases where (usually racist) vigilante violence had a huge impact on people's decision-making. But when the government does it it's different because people in government got there by proving their agency and ability to actually impact the world. Like, it feels almost like he's offended that the NPCs might try and do something as drastic as killing someone without GM permission.

Meanwhile in reality, people legitimately do feel like they don't have a lot of options to protect themselves from the real harms this industry is doing, to say nothing of the people who buy his line about the oncoming class-K end-of-life scenario. Anger is an appropriate response to the circumstances we find ourselves in, and in a nation that has been quietly cultivating a culture of heroic violence for decades we shouldn't be surprised to see people trying to inflict that fear and rage upon the outside world.

Maybe it's just because I'm rolling back through Age of Mythology, but I died laughing at "it's like the centaur, Helen"

I've never understood how these things are simultaneously gaining their abilities based on statistical analysis of all kinds of random writings online including social media, fanfic, reddit, etc. but also are simultaneously supposed to end up as experts rather than a much faster and more agreeable dumbass. Like, the training data may include all the great works of literature, all the scrapable scientific studies and textbooks they could steal, and so on. But it also included every moron who ever shared conspiracy theories on Twitter, every confident-sounding business idiot on LinkedIn, and every stupid word that Scott or Yud ever wrote. Surely the bullshit has to exceed the expertise by raw volume, and if they took the time and energy to curate it out the way they would need to to correct that they wouldn't be left with a large enough sample to actually scale off of.

Basically, either I'm dramatically misunderstanding something or the best we can hope for is the Average Joe on Reddit, who may not be a complete dumbass but definitely isn't a team of PhDs.

"Cursed spaghetti"

🤌

Sweden is an interesting example because they pioneered the let-it-rip approach to COVID. That was less disastrous than it could have been but not great even in a country with a lot of detached housing and nuclear families. https://kevinmd.com/2025/01/swedens-controversial-covid-19-strategy-lessons-from-higher-mortality-rates.html I would not have recommended putting children in daycare without strict indoor-air-quality standards between 2020 and 2024.

They do have consequences! Crowded schools and preschools and daycare spread all kinds of dangerous infectious diseases and there are consequences of getting them. With the arrival of COVID and the decline in vaccination risks are rising.

'Getting sick builds resistance' is another of the folk medicine beliefs which we in the infection control community have been fighting since 2020. Some diseases are milder the second or third time, but generally you want to get as few infectious diseases as possible.

the worst kind of violence, the sort against people like me

all those other deaths? those aren't violence

we also need to care more about property

The Zvi post really pisses me off for continuing to normalize Eliezer's comments (in a way that misrepresents the problems with them).

This happened quite a bit around Eliezer’s op-ed in Time in particular, usually in highly bad faith, and this continues even now, equating calls for government to enforce rules to threats of violence, and there are a number of other past cases with similar sets of facts.

Eliezer called for the government to drone strike data centers, even of foreign governments not signatories to international agreements, and even if doing so risked starting nuclear war.

Pacifism is at least a consistent position, but instead rationalists like Zvi want to simultaneously disown the radical actions, but legitimizes the US's shit show of a foreign policy.

Another thing that pisses me off is the ahistorical claim by rationalist that such actions are ineffective and unlikely to succeed. Asymmetric warfare and terrorist tactics have obtained success many times in history! The kkk successfully used terrorism to repress a population for a century. The black panthers got gun control passed in California and put pressure on political leaders to accept the more peaceful branch of the civil rights movement. The IRA got the Good Friday agreement. The US revolution! All the empires that have withdrawn from Afghanistan!

Overall though... I guess this is a case of two wrongs making a sorta right. They are dangerously wrong about AI doom, but at least they are also wrong about direct action and so usually won't take the actions implied by their beliefs. (But they are still, completely predictably, inspiring stochastic terrorists).

Also, it’s apparently enough for an LLM endpoint to be paired with an email inbox to be considered an “embodied agent”, words mean nothing.

This is a very interesting glimpse into the managerial class' psyche. A person is their email address. Very simple. Why would you need more than that.

It's ludicrous to pay taxes on the wealth your robots make, but it's savvy business to charge each software-delimited robot as a separate being - just like charging per-cpu-core was!

Ah right, I need to get a 365 license for word, which comes with a free copilot agent, who needs a 365 license for its copy of word, which comes with a free copilot agent, who needs a ...

JFC at least wait until you have a de-facto monopoly before musing about extracting the rents! This is capitalism 101.

Ahh sh*t if all my rent-seeking employee-reducing dreams come true, i'll lose money on my product subscriptions rents! Quick! I should come up with bullshit that will solve everything!

Do you get a refund when an "agent" inevitably blows out its context window and starts emitting deranged output, or does that automatically get rolled over into starting up the next "agent"

or is there more going on?

One idea I've read about (heavily developed by Ed Zitron, but also a few other news sources and commentators have put it forward) is that SaaS (Software as a Service) businesses were heavily over invested in expectation of basically infinite growth over the past decade. SaaS growth was "exponential" in its early days, but then various needs of the market were basically saturated, so SaaS companies squeezed more growth out cutting cuts or upping how much they charged, and now it is finally catching up to them.

The AI hype means almost everyone tries to interpret everything the lines of AI causing it. The recent price correction in many SaaS companies was (mis)interpreted as the threat of vibe-coded replacements forcing them to cut costs. The SaaS companies trying to cut costs and going through layoffs is being misinterpreted as AI successfully replacing junior devs.

that looks like a heaping pile of correlation, without mentioning the general downturn

So I don't have time to read the full paper and I probably don't have the background to make an informed critique of the methodology once I do (not that that's gonna stop me). But I feel like the challenge here is in mapping the distinction between junior and senior coding roles. To what extent do the senior coders get treated like a distinct job as opposed to being junior-but-seasoned?

Based on a quick amateur read of the abstract it looks like they're assuming the first option, that junior and senior developers are separate roles that can be largely disentangled. But if the other option is true, then in the event of a general industry downturn (say, after over hiring during recent periods of unsustainable growth) then it might make sense to look at the cuts to junior roles as simply removing the less efficient and effective people from the development role, rather than specifically cutting the juniors because they're uniquely exposed to AI replacement.

I don't know which model is more accurate to how the industry treats these roles or whether it varies by organization or what, but that's what seems like the most likely alternate explanation for the observed shift towards a very senior-heavy workforce.

And seniors obviously grow on senior trees (assuming that this take is actually true)

You've reminded me about the whole edifice of Qanon lore where they would try to combine 4chan (and later even sketchier sites like 8chan) hints with whatever Trump was posting at the moment to decode secret knowledge about stuff like when the military tribunals executing all the democrats would be.

Anyway, in Eliezer's case, I kind of get the feeling the lesswrong rationalists have somewhat moved on from him? They are still excessively deferential to him, but the vibe I get from hate-browsing lesswrong is that the majority of the rationalists there put much lower odds on AI doom? (It's hard to tell exactly because Eliezer has avoided committing to timelines or hard probabilities on AI doom despite all his talk about putting probabilities on everything in the sequences). Lesswrong occasionally references his tweets but not that often. Like I think sneerclub actually references them more often?

The optics is that his actions would paint EA in a bad light.

I think SBF's Scamfest Spectacular did a good enough job of that already :P

HN is more focused on the idea that Altman and co. are fomenting class hatred and the attack is more akin to Luigi Mangione’s attack on a health insurance CEO.

Altman and co's antics have repeatedly shafted the working class to billionaires' direct benefit, so I'll give the orange site that on the "fomenting class hatred" part. Thinking his actions are any way related to Luigi Mangione nailing the healthcare billionaire's fucking wild, though

writing this up for today, will be mentioning Ziz

more akin to Luigi Mangione’s

Didnt he also have sort of links to the LW idea space?