wzdd

Approximately 10 minutes is enough time to see the benefits but not the disadvantages. These effects might disappear again after multi-week interactions with AI, after it has led them down the wrong path, wasting hours; gaslit them into believing an incorrect solution is correct; or repeatedly misunderstood the problem.

This initial honeymoon phase happens whenever someone starts using a new tool, fades after weeks or months, and and is not the time to evaluate the deleterious effects of the tool.

They carry information in regular human communication, so I'm genuinely curious why you'd think they would not when an LLM outputs them as part of the process of responding to a message.

You can also configure the scroll direction on Windows (through the registry with 10, using the settings GUI with 11) and in Linux (depending your DE).

Kind of like what3words, except what3words uses three words which you stand a chance of remembering, whereas this produced, for an address similar to mine, "Miniature nerves eulogize gaily inside erect lion yet able stables hiss the conclusive consultation."

Advertisement, noun. A notice or announcement in a public medium promoting a product, service, or event or publicizing a job vacancy.

"DO NOT, EVER, UNDER ANY CIRCUMSTANCES, think of an elephant"

“In your repo” and “in the directory you are running copilot” are two separate things.

Fun. I did this recently with mine. There's now a discreet USB cable running down from behind it...

If comments here were for anybody to use as they wish, then anybody could use them for whatever they liked and (thus) YC could refuse to delete them. Being okay with both of those doesn't isn't a fundamentally-opposed viewpoint. One is a logical consequence of the other.

They've basically taken three separate models, one for fly vision, one for the fly brain, and one for the fly body, and bolted them together Frankenstein style.

They've taken the connectome, which is a map of how neurons in the brain are connected to each other, and then created a fly brain using artificial spiking neurons connected together using that same connectome.

So the neurons are not remotely accurate. The interesting point they're making is that even with these simplified neurons they still see plausible behaviours (i.e. simulate the presence of sugar by stimulating a gustatory neuron -> neurons associated with lowering the proboscis for feeding are triggered). So they make the case that a lot of information is encoded simply in the connectome.

The body isn't connected up to the brain in the way we'd expect. "Input" comes from a completely separate neural network which they've trained to simulate appropriate CNS neurons, and output is looking at "descending" (efferent I guess) neurons in a very basic way. It's not completely playing an animation, but the level of connectivity is very low dimensional. It's not clear how much control they have, but for example I imagine they have a spiking threshold for the proboscis below which it's lowered and above which it's raised, which is sort of like you being able to stick your tongue completely out or pull it completely in but nothing else.

So it's not especially bioplausible. The most interesting part is that leaky-integrate-and-fire connectome-based brain model which they're using, even though it is also very limited (for example, it doesn't learn).

Demo looks very cool though. Much credit for being so explicit about what's going on in it in that post. And I was immediately filled with ideas about what they could do next to improve it, which to me is a signal of good research.

He doesn't really even dig into the quality of Spotify's AI DJ apart from pointing out, in a very roundabout way, that it was designed for popular music.

Classical is a harder (or at least different) problem and it's why specialist apps like Apple Music Classical exist.

Actually, the article says the law says A or B. From the article:

> In a simplified form, retailers are responsible for warranty claims and must replace or refund the defective item

The choice is at the retailer’s discretion and the refund is obviously of the purchase price.

This is a non story. The retailer is acting lawfully. The only curiosity is that normally it is in a retailer’s best interests to replace goods, but in these circumstances they are better off refunding them.

I'm just annoyed I lost points for specifying both "Open the peanut butter jar" and "unscrew the lid of the peanut butter jar". The first one's context! The second "more precise" version doesn't specify what you should do with the now-unscrewed lid and the obvious solution, for a robot that takes things literally, is to leave it sitting on top of the jar.

Online tests have very explicit grading criteria which you can be confident are applied equally. AI evaluations, as the linked video in the article points out, not only do not have explicit grading criteria but the companies promoting them can't even describe what it is.

So the meaningful difference is that unlike a test you don't know what it's looking for and you don't know if it's ranking you objectively.

Over the top as mandatory registration of cyclists may be, he has a point on shared bike lanes. They're awful both for cyclists and pedestrians. From a cyclist's perspective you have a narrow path filled with effectively-immobile obstacles; from a pedestrian's you have a machine coming at you at speed which could cause you significant harm.

Not seeming unnatural is literally what the LLM is trained to be, but it's pretty interesting how little sense they make when you dig in. Goes to how little attention we pay normally, and/or how much weight we put on text seeming natural.

"A new lawsuit doesn’t just revisit the $40 billion Terra-Luna meltdown; it questions whether..." -- the purpose of a lawsuit is to question something (by making an allegation), you don't sue someone to "revisit".

"Ten minutes is not a coincidence. It is a trade." So is an hour, or thirty seconds, or...?

"Not just as bystanders, but as alleged participants" -- the "just" doesn't make sense; participants aren't bystanders.

Of the list, only "It reads less like a rescue offer" and "These are not isolated; they are part of Snyder’s broader efforts..." makes any sense in context.

The gotcha style "if AI is so good, why does $AI_COMPANY's software not meet my particular standard for quality" blog posts are already getting tedious.

The fact that different parts of the file use different endiannesses really added that special Apple tech flavour.

Currently all the comments are either talking about whether DisplayLink's emails were helpful or nit-picking the aesthetics of the site, so just to get technical...

While adding pause / resume functionality certainly solves the problem, it does seem like not the best possible solution. Firstly, it's a rather far-reaching change. Prior to this work TTY updates always succeeded, but now the kernel has to be aware that they can fail just in case userspace is talking to a DisplayLink (which the article acknowledges are increasingly rare)? We have a new type of wait queue (or wait type) for "waiting on TTY operation"?

Secondly, the parallel with serial (heh) links isn't great because with a serial link you have no idea what the other side is doing with the data, so you can't make any assumptions. But for a TTY you know its dimensions and furthermore you're the one doing the terminal emulation, so a) there's a bound (and quite a small one) on the amount of data you need to buffer and b) you know exactly how the data is going to be presented because you're the one doing the presentation. So there is an opportunity here to be more efficient than just forcing userspace to halt. A good analogy would be an X terminal emulator which faithfully draws every line of text, even if it's scrolling past hundreds of times faster than a human could read, versus one which updates its buffer as fast as possible, even if it's only redrawing at the display refresh rate -- the latter performs much better because it only shows the data that ultimately matters!

In particular, non-DisplayLink TTY drivers behave more like that performant X terminal emulator, because they're writing directly to graphics memory. Treating DisplayLink like a serial terminal makes it slower than it should be in the event of a lot of data being written; it is doing all its updates, even if they are immediately overwritten.

A more performant approach would be to store two text buffers, one for the current state (ie what DisplayLink is dipslaying) and another for the desired state. Diff the two to determine what to update when the DisplayLink is ready again.

It seems like this is basically what happens in graphics mode anyway, with dirty rects (which would just become larger dirty rects basically until the DL is ready for more commands) -- i.e. you have to buffer anyway for efficient blit / readback etc.

If diffing textmode feels too much like policy, make a user-space component. Or do what graphics mode does and use just one buffer with a bounded set of dirty rects.

In other words, it seems like a solution which came from "we are deep in the bowels of the device driver, what is simplest possible thing we can do?" and there's nothing wrong with that, but it does end up moving complexity elsewhere somewhat.

> In fact, I think the first photo used outpainting

And it looks bad. The streetlamps on the right are way too close together and the building on the left looks like a prison or something out of Dickens.

Python got this right. Zero-based indexing combined with half-open slice notation means as a practical matter you don't see too many -1s in the code. Certainly far fewer than when I wrote a game in Löve for a gamejam, where screen co-ordinates are naturally zero-indexed, which has implications for everything onscreen (tile indices, sprites, ...)

https://issuetracker.google.com/issues/371713238#comment829 seems fine and only has the potential to break other Apple BT hardware, which is relatively easy to test.

Nobody actually productively commenting in the thread thinks it's a conspiracy theory and everyone acknowledges that the Apple hardware is off-spec. It would be nice to see Android add this workaround.

The article states that Google has done the same for this extension as part of providing its "Featured" badge.

Last Chance to See has a fun bit about listening for dolphins in the Yangtze by taking a regular microphone and putting a condom over it. Always wondered how they sealed the end.

The wolf wasn't trying to be vegetarian, he was trying to not have the forest creatures run away from him. He achieved his goal by not eating them.

If you're 12 years old and it's the family computer (which I suspect was a not-insignificant part of the target market), you've also got to negotiate opening it up and sticking in a circuit board, adding drivers (and thus becoming prime suspect in every subsequent Windows crash), and finding some way of actually vibrating the air, which at the very least involves buying desk speakers and negotiating space on a desk which might not be your own.

Also you don't have $100.

Entertaining that the article about copyright-infringing similarity of AI-generated summaries is illustrated with a picture of an animated skeleton labelled "White Walker", which is neither what White Walkers are nor what they look like.

Looking forward to seeing the same generation of biohackers who previously CRISPRed their lactose intolerance to now CRISPR their hippocampi!

This explains a lot for me. On the server side, all my for-pay stuff is deployed using Docker. We have a single Python environment and complete control over it. We do multistage for compilation.

Client side, we don't get the privilege of deploying code: we need to build installers, which means again we have complete control over the environment because we package Python and all associated dependencies.

I'm sure there are marginal benefits to uv even with the above scenarios (better dependency management for example), but it seems that there's a middle ground here which I have largely avoided which is where uv really shines.

GP's point is that buring something down is by definition something that requires a specific physical process. It's not obvious that thinking is the same. So when someone says something like "just as a simulation of fire isn't the same as an actual fire (in a very important way!), a simulation of thinking isn't the same as actual thinking" they're arguing circularly, having already accepted their conclusion that both acts necessarily require a specific physical process. Daniel Dennett called this sort of argument an "intuition pump", which relies on a misleading but intuitive analogy to get you to accept an otherwise-difficult-to-prove conclusion.

To be fair to Searle, I don't think he advanced this as an agument, but more of an illustration of his belief that thinking was indeed a physical process specific to brains.