HNisCIS

I took a picture of a jpeg and compressed it as a jpeg for extra jpeg

Yes, but the point being for the LLM to be useful to us it has to do something relevant for us and we have to define what is relevant. Even if you view them as fully human level or beyond (in terms of agency) there's still some purpose that we have to help provide them with.

To put it differently, if you have some idealized model in front of you that can do anything a team of humans can do, what do you say to it? It's still just a model that takes an input and provides an output.

Maybe?

At the end of the day, even if they are some insane oracle (pun intended), they're still bounded by training data and how it relates to the real world. Even if they're a near perfect tool, we are still the interface between them and our lived experience. If that stops being the case then why do we care about the output?

This assumes it doesn't graduate to just killing all of us and doing it's own thing, but within this paradigm it doesn't really have goals.

You develop a set of heuristics for skimming it in the same way you do with code.

Big wrapping operations like sums, integrals, and matrices, then what's nearby them, give you a very good idea of where things are going context wise.

FWIW I quite liked it but I'm from the RF world so maybe I'm biased.

It's a 4th power because you're doing the inverse square law twice, once for the outbound pulse and once for the reflection.

8B is still three orders smaller than current frontier though

The anti counterfeit strategy is that the volumes here aren't worth it for someone to clone, even if the board and code are dead simple.

These niche hobby products are usually reasonably safe because they rely on community network effects, becoming the darling of a random subreddit or discord group and people aren't going to save $20 to get some device of dubious origin.

Still a strawmam somehow?

I fear you underestimate the malice of the average tech company

Medical/psychological experiments involving human subjects are mandated by federal law to run through a review board. Did that happen here? This isn't a medical experiment, but I don't think its nuts to feel like maybe there should have been some oversight.

That's not what the parent is talking about. You're using the strawman of what appears to be occupancy zoning or something similar to argue that a techno fascists should have free reign to manipulate the mass movements of people within a city for their own purposes.

Yeah, it always strikes me when this crazy, obviously stupid stuff ends up "49 to 51" like either there are active forces fighting each other ending in an equilibrium, or the equilibrium is arising passively, everyone and their representatives are just randomly scribbling in the circles.

Idk if many people remember how Reddit was when it started, its whole draw was that it was a bunch of relatively intelligent/friendly western internet people who liked to kinda role play as a hive mind.

I wouldn't call it the most censored platform, but it still has vestiges of its old self, particularly with respect to philosophy and politics.

I see this parroted constantly but it seems to be getting proven wrong by Japan and, frankly, the US

I agree with the second part, 90% of apps are just a vague rehashing of 100 other apps, very little novel software ever gets written, especially at the component level. The point being that if you want to create truly novel systems you best know how to program, even if an LLM is writing the less novel parts of that system.

If we're smart about this, maybe it means we get to do novel work a little more frequently. That said, I fear that a lot of people don't look at every single LLM output and think "Eh it's workable but it doesn't spark joy" and anyone who doesn't think that is likely going to be seeing their QOL decreasing. At that point your time might be better spent learning how to make Molotovs.

Here's a much simpler reason: the output of an LLM, mathematically, is average. The code it writes is, effectively, the geometric mean of github. The ideas it comes up with are the geometric mean of wikipedia and reddit. Go ask it how to do something truly "out there" and watch it give you the most milquetoast reply you've ever seen. I asked Fable 5 to create a profitable business and the best it could come up with was:

Sell an expensive, recurring B2B service into your network, fulfill it with technical leverage, then productize it into software once you've done the same work five times.

Yeah, good fucking work, nobody's ever thought of that. It's obvious why it picked that though, every 3rd post on this forum and half of the rest of the internet is singing the praises of be-your-own-boss SaaS passive income and it turns out that if you convolve all of that into a black box this is what you're gonna get.

I take the opposite view but to the same conclusion. I've heard the joke "the more advanced you get the worse you get at math" from a lot of mathematicians. I'd bet most mathematicians can't do a long division problem without having to think pretty hard about it, because they've long since moved past that skill.

All that said, people still learn and do manual math at all levels in order to advance the field even though calculators and python notebooks exist.

I think people get too wrapped up in LLMs being the entirety of the future when the LLMs themselves are entirely a product of the past. An LLM is less a thinking machine than a lossy JPEG for language and to an extent, knowledge represented in language. As such, if you want to expand the field and move into the future you aren't going to rely solely on an algorithm that reverts to the mean.

The manufacturing is really deeply gnarly. Like crazy materials you'd have to reverse engineer the production process for, then reverse engineer how to form the blades so they'll survive insane loads for thousands of hours of operation without failure under extreme temperatures. It's like saying you can make a MacBook chassis because you have a block of aluminum, some hand files, and an Adderall prescription.

It's actually safer-ish. First: terms. Filtering is good, involves moving slowly through stopped traffic between the cars, usually under 20mph differential.

Splitting is less good, that involves weaving between cars at speed and is actually dangerous.

Some of the worst accidents are rear endings where drivers (not paying attention) just run you over while stopped in traffic.

This is offset by accidents where people do the stuff you're worried about but when it's practiced correctly that's not as big of a risk and generally leads to less catastrophic accidents than rear endings.

It's also just kinda dumb to force a class of vehicle that can get out of traffic jams to instead sit in them

Motorcycles aren't invulnerable 3 ton steel tanks but the stats and anecdotes are deceptive. They're really not that bad if you're not a moron, even if you're mostly worried about other road users. The stat are wildly bimodal.

~30% of deaths involve drunk riding

~30% of deaths involve not wearing any helmet (let alone full face ECE 22.06 rated ones or any other gear at all)

~30% of deaths involve someone with no motorcycle licence.

These aren't all mutually exclusive obviously, rather the Venn diagram probably looks rather...circular.

The issue isn't so much everyone trying to kill you, you can fix a lot of the visibility issues and you have some additional options if someone is about to hit you. The problem is that two wheels make for a VERY dynamic system and you're managing two different brakes with weight shifting between two wheels based on your inputs. To that end ABS and TCS are absolutely huge, IIRC something like >60% safety improvement.

Tldr don't buy an old retro bike with no safety systems and ride it drunk without a license or gear, you'll continue to pad the numbers.

It's so fucking bad. I'm watching a team try to maintain a huge dashboard/control application that interfaces with a large amount of hardware using solely AI workflows.

Literally nothing works, all the timers/time counters are different across the pages, constantly commands hardware to do stupid shit, breaks during critical moments/in front of clients.

Eventually mgmt had to institute change freezes for high profile events because the team was breaking too much shit all the time.

The average C suite dipshit doesn't realize that the performance drops off a cliff once your project is more than some fraction of the context window so they will make pretty dashboards all day long but once you need to cover all the edge cases of a real system it all explodes.

AI isn't trained on the type of software style we'll need to create systems using AI, it's trained on how we used to write software. It doesn't reuse code or elegantly structure annoying, it just adds more code until the thing builds and passes some fake tests, even if half of it is functionally dead/unused.

The issue wasn't really git master, that was a side effect. Google what MISO means and maybe go learn some actual kernel development.

On one hand yeah it's a little over the top.

On the other hand I don't think this was written for our proto-techno-fascist forum...

There are a multitude of ways to significantly curtail crime that don't rely on this paradigm of spying on everyone. That's like saying "I can't get to work on time, we need to keep making the highway wider".

I don't think you're wrong broadly, though I want to add that the particular interceptor warheads were relatively small and nukes detonated in the upper atmosphere generally don't do much to stuff on the ground. There's not enough material to create significant fallout so it's just a mild EMP and even milder pressure wave

Totally different situation. People are removing those words as a sign of respect and a very small number of people are chasing down those that don't because it implies an open lack of respect.

Two different discussions, but I've had an earthy crunchy employer ask me to put in for one once.

For pure generic full-stack-whatever devs yes. For EEs, embedded, FPGA, RF, etc you can pull waaaaay more in the defense world, especially if you're willing to do cleared work.

I can't name names but 3 of the startups I've worked at.

Places I haven't worked:

Skydio

Applied Intuition

Saildrone

Planet Labs

Boom

Scale AI

Also worth noting that sometimes it's on purpose, sometimes the founders are all "we're gonna save the world" then AFWERX enters the chat with a big fucking check and the founders yell "Nevermind! Guess we're the baddies now! How many slaughterbots did you say?"