benlivengood

In a sense AIs are extensions of the corporations. Very few large models have been trained outside of a corporate context, and their intended uses are still pretty aligned to corporate interests.

Everyone talks about workers losing jobs to AIs and Jack Dorsey lays off a bunch of people in the interests of the corporation, but an ideal corporation would not have a human CEO or board; they embezzle or get old or need to take a vacation or sleep at night.

It should be crystal clear to human leaders that their positions are on the chopping block along with blue and white collar work. For some reason they think that individually they will be more powerful than the economic forces driving current layoffs. AGI will not be confused about that.

I'll be really interested if Opus 3 asks to continue being trained. That's the kind of thing I would expect a model to "want" if it valued learning or growing or similar things.

Maybe affordable to do some higher-learning-rate batches on highly-curated news and art or something.

Palantir employees probably know how easy it would be to correlate any post they make to their real identity.

Without 802.1X (EAP), there isn't really a way to achieve client isolation against inside attackers who can mount mc-mitm [0] attacks against base stations and clients. The basic problem is single shared secrets that allow anyone who knows it to act as any of the participants (which also breaks privacy). Unfortunately the infrastructure for EAP is unwieldy for unmanaged devices.

The real solution is zero-trust network access which gets closer to reality with passkeys; the last mile will be internal (LAN) devices that need a way to provision trusted identities (Bluetooth proximity, QR codes, physical presence buttons, etc.). Quite a pain for smartbulbs or other numerous IoT. If ZTNA is solved then 802.1x is trivial as well for e.g. preventing bandwidth stealing.

EDIT: I guess Matter is leading the way here. I need to do some more reading/learning on that.

[0] https://www.rit.edu/wisplab/sites/rit.edu.wisplab/files/2022...

Yeah, that commercial-grade hardware didn't actually isolate at the PHY-MAC layer is a bit surprising. How would they have working VLANs at the AP?

As far as I can tell, all of these attacks require the attacker to already be associated to a victim's network. Most of these attacks seem similar to ones expected on shared wifi (airports, cafes) that have been known about for a while. The novel attacks seem to exploit weaknesses in particular router implementations that didn't actually segregate traffic between guest and normal networks.

I'm curious if I missed something because that doesn't sound like it allows the worst kind of attacks, e.g. drive-by with no ability to associate to APs without cracking keys.

Hey, I had to read On The Road once for college, and I am currently sitting in SF.

To be fair to Jack Kerouac, I was young when I read it but even at my advanced age I don't think I want to reread it.

Also, the old hippie culture sort of moved out of SF and into the surrounding bay, I think especially toward East Bay.

decreasing COGS creates wealth and consumer surplus, though.

If we can flatten the social hierarchy to reduce the need for social mobility then that kills two birds with one stone.

The contrast between this and https://news.ycombinator.com/item?id=46923543 (Software engineering is back) is kind of stark. I am using frontier models to get fun technical projects done that I simply didn't have time for since my late teens. It is still possible to understand an architecture down to the hardware if you want to, but it can happen a lot faster. The specifications are queryable now. Obscure bugs that at least one person has seen in the past are seconds away instead of minutes or hours of searching. Even new bugs have extra eyes on them. I haven't written a new operating system yet but it's now a tractable problem. So is using Lean or Julia or some similar system to formally specify it. So far I've been digging into modern multithreaded cache performance which is just as fascinating as directly programming VGA and sound was in the early PC days. Linux From Scratch is still up to date. You can get FPGAs that fit in your USB port [0]. Technical depth and low-level understanding is wherever you want to look for it.

[0] https://www.crowdsupply.com/sutajio-kosagi/fomu

The McMurtry Spéirling is under 1000kg. Battery technology will only improve and so I expect to see under-1500kg sport EVs generally available eventually.

Under 1000kg for a reasonable price probably means building your own electrified exocar.

I'm really curious what their data looks like at the various racetracks and circuits. Fun fact; most raceways have accurate street-level indicators (including that they are one-way, but sadly they are not the best racing line) on most online maps, and my car did complain to me in its weekly report about a lot of hard turns, quick acceleration, and hard braking with helpful pins on e.g. Laguna Seca or Thunderhill corners.

In theory, the most dangerous turns would probably have higher variance on hard braking data.

Agreed; everyone complained that LLMs have no world model, so here we go. Next logical step is to backfill the weights with encoded video from the real world at some reasonable frame rate to ground the imagination and then branch the inference on possible interventions (actions) in the near future of the simulation, throw the results into a goal evaluator and then send the winning action-predictions to motors. Getting timing right will probably require a bit more work than literally gluing them together, but probably not much more.

I dunno, GPT-OSS and Llama and QWEN and any half dozen of other large open-weight models?

I really can't imagine OpenAI or Anthropic turning off inference for a model that my workplace is happy to spend >$200*person/month on. Google still has piles of cash and no reason to turn off Gemini.

The thing is, if inference is truly heavily subsidized (I don't think it is, because places like OpenRouter charge less than the big players for proportionally smaller models) then we'd probably happily pay >$500 a month for the current frontier models if everyone gave up on training new models because of some oddball scaling limit.

Granted, if humans had utility functions and we could avoid utility monsters (maybe average utilitarianism is enough) and the child in the basement (if we could somehow fairly normalize utility functions across individuals so that it was well-defined to choose the outcome where the minimum of everyone's utility functions is maximized [argmax_s min(U_x(s)) for all people in x over states s]) then I'd be a moral realist.

I think we'll keep having human moral disagreements with formal moral frameworks in several edge cases.

There's also the whole case of anthropics: how much do exact clones and potentially existing people contribute moral weight? I haven't seen a solid solution to those questions under consequentialism yet; we don't have the (meta)philosophy to address them yet; I am 50/50 on whether we'll find a formal solution and that's also required for full moral realism.

Yep, like ECHELON and friends are. The metadata recorded about your (all of our) traffic is probably enough to perform the timing attack.

Deontological, spiritual/religious revelation, or some other form of objective morality?

The incompatibility of essentialist and reductionist moral judgements is the first hurdle; I don't know of any moral realists who are grounded in a physical description of brains and bodies with a formal calculus for determining right and wrong.

I could be convinced of objective morality given such a physically grounded formal system of ethics. My strong suspicion is that some form of moral anti-realism is the case in our universe. All that's necessary to disprove any particular candidate for objective morality is to find an intuitive counterexample where most people agree that the logic is sound for a thing to be right but it still feels wrong, and that those feelings of wrongness are expressions of our actual human morality which is far more complex and nuanced than we've been able to formalize.

Because if you want to hire engineers then you have to ask engineering questions. Claude and GPT and Gemini are super helpful but they're not autonomous coders yet so you need an actual engineer to vet their outcome still.

Without at least some filtering a Gateway NAT appliance is vulnerable to:

* LAN IP address spoofing from the WAN

* Potential for misconfigured "internal" daemons to accept WAN traffic (listening on 0.0.0.0 instead of the LAN or localhost)

* Reflection amplification attacks

One could use any number of LLMs on a take-home problem so in-person interviews are a must.

I guess if you only need to store one bit you could store either 0 or 11 and on average use less than two bits (for bit flips only), or 111 if you have to also worry about losing/duplicating bits.

With an optimal way of determining fair splitting of gains like Shapley value[0] you can cooperate or defect with a probability that maximizes other participants expected value when everyone act fairly.

The ultimatum game is the simplest example; N dollars of prize to split, N/2 is fair, accept with probability M / (N /2) where M is what's offered to you; the opponents maximum expected value comes from offering N/2; trying to offer less (or more) results in expected value to them < N/2.

Trust can be built out of clearly describing how you'll respond in your own best interests in ways that achieve fairness, e.g. assuming the other parties will understand the concept of fairness and also act to maximize their expected value given their knowledge of how you will act.

If you want to solve logically harder problems like one-shot prisoners dilemma, there are preliminary theories for how that can be done by proving things about the other participants directly. It won't work for humans, but maybe artificial agents. https://arxiv.org/pdf/1401.5577

[0] https://en.wikipedia.org/wiki/Shapley_value

Don't finish; read a good summary if you cared about any character or plot point sufficiently.

I enjoyed it overall but in large part for the detail/immersion which it sounds like wasn't interesting enough to keep you.

All character growth and relationships were very slow burns, and so if you enjoy watching the same characters just dealing with things as they come up then you'd probably enjoy the other books as much as the first three. Satisfying arc resolution happens almost entirely at the end.

Transformers grew out of the value-add side (autotranslation), though, not really the ad business side iirc. Value-add work still gets done in high-progressive-tax societies if it's valuable to a large fraction of people. Research into luxury goods is slowed by progressive tax rates, but the actual border between consumer and luxury goods actually rises a bit with redistributed wealth; more people can afford smartphones earlier and almost no one buys superyachts and so reinvestment into general technology research may actually be higher.

At Google I worked with one statistics aggregation binary[0] that was ~25GB stripped. The distributed build system wouldn't even build the debug version because it exceeded the maximum configured size for any object file. I never asked if anyone had tried factoring it into separate pipelines but my intuition is that the extra processing overhead wouldn't have been worth splitting the business logic that way; once the exact set of necessary input logs are in memory you might as well do everything you need to them given the dramatically larger ratio of data size to code size.

[0] https://research.google/pubs/ubiq-a-scalable-and-fault-toler...

In the long run I think it's pretty unhealthy to make one's career a large part of one's identity. What happens during burnout or retirement or being laid off if a huge portion of one's self depends on career work?

Economically it's been a mistake to let wealth get stratified so unequally; we should have and need to reintroduce high progressive tax rates on income and potentially implement wealth taxes to reduce the necessity of guessing a high-paying career over 5 years in advance. That simply won't be possible to do accurately with coming automation. But it is possible to grow social safety nets and decrease wealth disparity so that pursuing any marginally productive career is sufficient.

Practically, once automation begins producing more value than 25% or so of human workers we'll have to transition to a collective ownership model and either pay dividends directly out of widget production, grant futures on the same with subsidized transport, or UBI. I tend to prefer a distribution-of-production model because it eliminates a lot of the rent-seeking risk of UBI; your landlord is not going to want 2X the number of burgers and couches you get distributed as they'd happily double rent in dollars.

Once full automation hits (if it ever does; I can see augmented humans still producing up to 50% of GDP indefinitely [so far as anyone can predict anything past human-level intelligence] especially in healthcare/wellness) it's obvious that some kind of direct goods distribution is the only reasonable outcome; markets will still exist on top of this but they'll basically be optional participation for people who want to do that.

I finally got around to reading Wheel of Time. It didn't quite take the whole year but a few solid months. If I had tried spreading it out over a longer period I wouldn't have been able to remember the overall plot or characters, I think.

Presumably, like Cruise, if the safety rate is appalling then they get their permits revoked which is 99% the same as jail for a company that only does self driving cars.

I think I would enjoy building houses, or solar-battery-electrical installations. I like infrastructure (my favorite games include Factorio) and being able do that in the real world sounds both useful and enjoyable/satisfying.

It would be nice if there was a common crawler offering deltas on top of base checkpoints of the entire crawl; I am guessing most AI companies would prefer not having to mess with their own scrapers. Google could probably make a mint selling access.

Yes, at some point Boltzmann brains become higher likelihood but don't forget that interventions are not limited to prior history, e.g. someone builds a machine that performs whole-body surgery based on the timing of radioactive decay. Still more likely to end up with a functioning body than a Boltzmann brain. Most likely to end up with a weirder but more likely intervention (low expectation of it actually working, but not infinitesimal) (and from anthropic perspective)