phdelightful

If you can tolerate a .gov email address, there are high-performance computing, edge computing, and AI groups at DOE labs. My impression is that they're always starved for talent in these areas due to stiff competition from industry and lifestyle restrictions that come with a security clearance (being a US citizen, reduced international travel, no drugs other than alcohol and nicotine, your work may require on-site). Defense labs are probably similar.

People who feel strongly that current procurement strategies are the wrong way to go about things should devote as much toil and treasure as they can to seeing their preferred solution materialize. In the mean time, I hope they will allow others to work within our existing system to keep our adversaries from replacing the US with something much worse for the entire world.

I basically agree. I don't think the political mechanisms exist in the short-to-intermediate term to get salaries in government labs high enough. As you point out, the extent to which "civilians" will be willing to participate in these lines of work is an open question (said without judgement, just as an observation). Presumably folks who have opted-in to work at defense contractors will be more willing, but do they have the ability to attract a sufficient amount of talent in this area?

Can the US taxpayer tolerate (financially, politically) what it will take to get the best and brightest to do AI work for .gov and .mil instead of adtech? All you have to do is look at defense lab salaries to see there is going to be a problem keeping talent around. Some people are certainly willing to take a pay cut to work on national defense, but probably not enough. Perhaps especially not the kind of person who went into adtech in the first place.

This outcome from using a large language model to mimic reasoning isn’t surprising. What’s surprising is Yan LeCun’s childish and petty reaction to this entirely foreseeable series of events:

> Galactica demo is off line for now. It’s no longer possible to have some fun by casually misusing it. Happy?

He’s supposedly an expert in this sort of thing

You don’t have to worry. There are physiological responses to pain and trauma that the anesthesiologist can observe, even if you are unconscious. Things as simple as heart rate and respiration all the way up to brain wave monitoring. This is part of the anesthesiologists job.

As an undergrad I made a guitar tuner using a PIC32 microcontroller to do time-series correlation to figure out the frequency (FFT was too slow for real-time) and a state machine embodied in a stratix FPGA to control a stepper motor to turn the pegs. A different era of “mobile app” I guess.

Not the parent, but I like it!

First sentence is unnecessary.

It would be better if it disallowed taking data for a legit reason and then using it for another less scrupulous reason. Point 5 may mitigate this concern substantially however.

Nice idea, but our current world except without US nukes is a much worse world. We’re sort of stuck in a local minimum of nuclear weapon danger for the time being.

Also I think the main accident in the article was during research into converting enriched uranium for civilian use.

Between OpenMP (older computers), CUDA (Nvidia), HIP (AMD), now SYCL (Intel), not to mention more interesting architectures, DOE is probably plently busy and has plenty of very fast computers that are hard enough to keep their codes working on. Aurora being delayed has damaged Intel's reputation and gets chuckled about, but I'm not sure how big of an effect it's really had on the ground.

If you want to see this line of thinking taken a bit too far, check out the list of Trilinos packages on github: https://github.com/trilinos/Trilinos/tree/master/packages

There's ~50, and nearly all of them are incomprehensible. It definitely makes things much less accessible to a newcomer / outsider.

(Trilinos is a set of scientific / engineering libraries for HPC)

> you cannot reason someone out of a position they have not been reasoned into

Yes you can, that’s one of the best things about reason. I see this phrase pop up from time to time and I guess it’s just too cute to die

Only if you also include instructions to decode it using the ten hundred most common words, I think!

I wish those people over 40 hadn't put a US government in place that abolished the fairness doctrine. Then our news sources would probably not be so bananas.

I like “from A to B” because B happens after A temporally and in the text. It also serves to emphasize A as opposed to B, which is nice because we usually wouldn’t leave A unless there was really a problem with it even if B is better in some way.

Maybe the positivists’ antipositivism should just be “not everything demands an empirical explanation in order for us to continue functioning well in day-to-day life, and that’s just fine.”

If you're asking why banks lend money to each other, it's mostly because banks need liquidity (e.g. $ in an account) in order to do transactions and meet legally mandated reserve requirements. If I'm a bank, rather than constantly be sitting on a huge pile of money for the (relatively) rare event that I have a huge transactions to make, I'll mostly do something more useful with my money, and then borrow it from another bank on a short-term basis when I need a bunch of cash.

The fed does this for banks that can't find another bank to transact with, albeit at a slightly higher rate. The fed will also borrow money from banks at a slightly lower rate. This causes the banks to naturally lend among each other between those two rates. This is how the fed controls the interest rate.

The other thing the fed does is buy treasuries. I think this is where the "printing" of money happens. The fed buys a treasury on the market and pays for it by printing money and placing it in the appropriate account. This is how the fed controls money supply.

I take your point. I think I was just reacting to various flavors of “he’s not crazy” as having a potentially normalizing effect but that may not be what was intended, or what other readers focused on.

I know cold turkey works for some people, but I’d never recommend it unless it’s truly an emergency (immediate danger of harm) and you have a support system around you to hold you to it.

The truth is, people are obstinate about change and also infinitely adaptable. So, make the changes small so you can get over the hump, and then stick with it ferociously. In 2 weeks it’ll be second nature.

You gotta have something fulfilling in your life away from a screen. Cooking, pets, exercise, relationships, kids. It may seem impossible to get there from where you are, I’d start with exercise. Great thing about exercise is it’s entirely fungible and each incremental increase is good. Start with anything and do it /before/ you do technology. Then don’t worry about it again until tomorrow.

Don’t you have to be at least a little bit more detached from reality than the median person (“crazy”) to shoot up a supermarket?

I don’t think crazy in a blank check to get away with whatever you want, you can definitely make yourself crazy in entirely foreseeable and avoidable ways that are your responsibility. This can still render you unreliable when it comes to your own motivations.

I want to see a fully-synthetic multimodal social media influencer that is nearly indistinguishable from reality. She does the same thing as the real ones except everything is completely artificially generated (housing, clothes, vacations, social circle). All text/image/video posts are completely synthetic but internally-consistent with this fabricated persistent universe. The only real things would probably be product placement. If you’re a brand, you’d just make a new online influencer instead of finding an organic one.

Is there a C++ library that implements static integer types with these ideas? In principle the operations don’t seem complicated, but there are probably enough edge cases that it’s tricky to get it all right.

The word “formula” doesn’t appear in the article you linked. It seems to be about other baby foods.

As an executive member of a nonprofit, this was exactly the kind of headache I hoped for. You can do as lot of good with $100K. Definitely worth the trouble.

> In justified text, it divides into appropriate syllables a word that lands on a line break, a task that machines have not yet mastered

Maybe I’m misunderstanding but Latex does this pretty well (“mastered” is a bit fuzzy I guess).

I'd remove the collective action problem. That would empower humans to solve their big problems as they see fit!

The counterpoint is that variable-length decoding introduces sequential dependence in the decoding, i.e. you don’t know where instruction 2 starts until you’ve decoded instruction 1. This probably limits how many decoders you can have. If you know all your instructions are 4B you can basically decode as many as you want in parallel.

I wonder to what extent this is an early step toward walling off Reddit's content from third-party search engines. They probably recognize that without a functional search they'd take a big hit in such a scenario, but with search under their own control they can better influence how people end up seeing various content.

I'm currently in the "smart TV but don't connect it to the internet" crowd. I think there are three broad criticisms.

Violating your privacy in unsurprising ways that provide no value to you: obviously when and how you use it, possibly including screenshots and audio fragments to help HQ determine what you're watching

Degrading your experience: basically just ads in places you don't want them that didn't exist when you bought it.

Violating your privacy in "surprising" ways that provide no value to you, and circumventing your efforts to stop it: what other devices are present on your home network, where your home is, connecting to unsecured wifi (or soon mesh networks with your neighbor's TV) to do so.

IMO these are the things to defend against. If they're not happening to you then that's great. It's also fine if they don't bother you. For example, if 3 is happening to me I'm not going to go to the trouble of figuring it out how to stop it, though it will change my future purchasing decisions.

Since you don’t offer much to work with, I assume you mean:

“I disgree. People at the cutting edge should be educating others both to help them make their ideas accessible and connect them with the future of their discipline.”

I encourage you to add more than a single word to the conversation.