The embedded programs can be connected to the other weights during training, in whatever way the training process finds useful. It doesn't just have to be arithmetic calculation. You can put any hard-coded algorithm in there, make the weights for that algorithm static, and let the training process figure out how to connect the other trillion weights to it.
I don't think anyone is using LLMs for those conversations. A lot of those replies are bots. There's a market for reddit accounts that have a solid human-looking reply/post history, to be used for astroturf marketing, so some organizations set up bots to grow such accounts. There probably are also just people who overuse "Honestly? [statement]" sentences. I've spoken to such people in person before LLMs.
> borrowing ordinary mannerisms of speech that aren't necessarily egregious
That's how a trope starts. When a minority of writers are using a particular pattern, it's personalized style. When a majority of writers in a genre adopt the same personalized style, it's a trope.
We find AI tropes especially annoying because there are three frontier LLMs producing a sizable chunk of text we read (maybe even a majority of text, for some people) lately. It would also be annoying if a clique of three humans were producing most of the text we read; we'd start to find their personal styles annoying and overdone. Even before LLMs, that was a thing that happened in some "slop" fiction genres where a particularly active author would churn out dozens of novels per year in one style (often via ghostwriters, but still with a single style and repetitive plot pattern).
No it's not. It's based on how much he "made" in the first half of 2020, mostly originating from gains in Amazon's stock, in a period specifically selected to inflate the number. If you actually want to display how much Bezos made since the user opened the page, there are many public APIs to get live stock data and you could show the actual live gain/loss. But that wouldn't really support the point you're trying to make, since there would be days where he actually loses more money than most people ever see.
He kept 80%. The other 20% is owned by 8 different VCs. Seems like he's still in control. There's value in using other people's money instead of your own because it might make him less emotionally risk-averse in how he manages it.
Do you mean the value in this specific tool, or in the concept? You don't need a dedicated tool to store agent session transcripts and link them to commits. This can be accomplished by a 10-line bash script.
Would declining to let you through security actually be a "penalty" (legally speaking), though? There are a ton of things you need to show papers for in the US; I can't imagine that all of them were pre-approved by the OMB.
How does that make it different? More features could always be added to most buildings. You could keep adding rooms onto the side, update the floors/ceilings/walls every year to stay trendy, add a water feature, expand the basement with a tunnel network, etc.
I find it interesting how people say "The US" to refer to groups under the US government that are often completely at odds with the interests of the actual US public. There are virtually no Americans who want our government to be acting in the interests of arms manufacturers except the arms manufacturers themselves and the politicians they pay.
Throughput would still remain unchanged. Suppose that the "lunch rush" is from 11AM to 1PM, and imagine that it's uniform for the sake of simplicity. Then drivers would end up being fully utilized from 11:10AM to 1:10PM, instead of 11 to 1. The 10 minute lag at the end where drivers are still finishing the queue makes up for the 10 minute delay at the start.
That's because the unannounced firedrills don't involve setting the building on fire. A "drill" equivalent would be if we all pretended the internet is down sometimes, and in some cases that still might be impossible to do without negative consequences.
You don't really need to win an argument with luddites. Completely rejecting extremely useful technology and then picking a fight with people who don't is a way to speedrun losing, whether you have "compelling arguments" or not. If the Luddites were correct, they wouldn't be dead.
> Ok, if that's really your thinking then you need to lay out: here's an impossible-to-ignore thing we can do with this, and this is how, and this is why this wouldn't be possible without this thing.
There was a period of ~1000 years where you could also make this argument against some high-minded guy advocating for democracy.
I believe you're misunderstanding what the OP means about "long-term" memory. From what I can tell, it's not actively modifying the weights of the underlying model, it just "remembers" things from a high number of tokens into the past of its context. The point is that this allows it to remember something it read ~200 pages ago in a very long context window, not that it can remember something from one session into another clean session.
Signing something doesn't verify that it's real, it just verifies that you claimed that it was real, which everyone was already aware of. You can either hack a camera, or use an unhacked camera to take a picture of a fake picture.
Suppose that I care about trustworthy and reliably accurate news sources and am willing to pay. How can I distinguish which ones are trustworthy and reliable? No offense to the folks at 404 Media, but I've never met a single one of them, and I have no reason to believe that they wouldn't lie to me for money. You clearly have your own prejudices and biases about which media organizations are honorable and which are not, which you're wrapping up as if it's about a "truthfulness" that you couldn't possibly actually verify.
Would calling and saying, "Hey, the bridge is destroyed!" without an image not have also triggered a delay? I question the safety standards of the railway if they would just ignore such a call after an earthquake. Generative AI doesn't change the situation at all. An image shouldn't be treated as carrying more weight than a statement, but the statement without the image would be the same in this situation. This has really been an issue since the popularization of the telephone, which made it sufficiently easy to communicate a lie from far away that someone might choose to do so for fun.
Any attempt to fix this would immediately used for outright immigration fraud, and any attempt to ban immigration fraud will always be used as a cudgel for bureaucrats to keep out normal, tax-paying Europeans, until low-level bureaucrats have term limits.
When someone is a danger to innocent people walking by who didn't choose to do any fentanyl, their recovery chances are secondary to the safety of the innocent passers by. The people who advocate for leaving them on the street never want to take responsibility when one of them kills a random kid for fun. That may be something that only a small minority of fentanyl addicts are going to do, but it's not something that we have any obligation to allow in the name of helping drug addicts.
Luxembourg has insane tax revenue per capita because of its status as an international tax haven. A program that might be hardly noticeable on Luxembourg's budget could put a big dent into the budget of an American city.
Good luck getting Congress to do anything at all, ever. Best we can offer you is one big partisan pork bill every two years, no matter who's in charge and no matter what voters actually want. Maybe if one senator (foolishly) decided to make penny rounding his pet issue, and worked his ass off at it, he could get half of it wrapped into the bi-annual pork bill.
That license also doesn't give the ffmpeg devs the right to dictate which bugs you're allowed to find, disclose privately, or disclose publicly. The software is provided as-is, without warranty, and I can do what I want with it, including reporting bugs. The ffmpeg devs can simply not read the bug reports, if they hate bug reports so much.
I see people say this literally every time someone complaining about lack of interviews posts their resume. We shouldn't have a system where every job seeker is supposed to be more of a resume formatting expert than the average HR rep. The fact that someone looking to hire is going to see an okay resume of a highly qualified candidate and say, "LOL too long; didn't read" is the most glaring symptom of what he's talking about.
> Maybe a site like this should try to root-out the "root source" of information - official press releases or press conferences, eye-witness accounts.
I'd personally pay for that. It feels like 90% of the "news" I see these days is just some site telephoning what a different reporter said. I regularly see "study finds x" articles that completely bury the original academic source. Often, "politician said x" articles that spend a lot of time going over everyone's reactions to whatever the politician said without letting me have the full video or press release where he actually said it.
This would also fix an issue I'm seeing on Kite where some stories seem to be the same thing from different angles (one article about the Texas floods is directly above an article about how the Texas floods are "testing FEMA", and there are two separate articles for the recent Trump-Netanyahu deal, one in World and one in USA. X's CEO resigning is three different articles in different feeds. The Business tab has two different articles about Trump tariffs that could really be one article).
> hallucinate whole APIs for D3 on multiple occasions, which should be really well represented in the training sets
With many existing systems, you can pull documentation into context pretty quickly to prevent the hallucination of APIs. In the near future it's obvious how that could be done automatically. I put my engine on the ground, ran it and it didn't even go anywhere; Ford will never beat horses.
> In 1930 Keynes projected that by 2000 their grandchildren could work 15 hours per week
This projection came true. You can live with a 1930s living standard while working 15 hours per week at a median wage. However, nobody wants to live at a 1930s living standard. It turns out that in aggregate people prefer to spend productivity gains on conveniences and higher standards of living than on working fewer hours.