skybrian

10 ms·

241.
by skybrian 2mo ago
You tell the AI what broke and it fixes it, and this is usually a better experience than contacting support. But what if that fails? This might not generalize beyond developers, or at least not right away. Even if we’re not doing the upgrad
242.
by skybrian 2mo ago
If you don't try to have fun with it then you're not going to have fun.
243.
by skybrian 2mo ago
Huh, I thought I was reasonably good at this but didn't see them. Rather than getting annoyed about it, maybe it could be a fun puzzle? Spot the AI glitches in this image. Maybe do that contest before giving the award, though, among th
244.
by skybrian 2mo ago
Seems like ideas are not really free if you can't use them to build your own theme park? But it's harder to get anyone to visit.
245.
by skybrian 2mo ago
Still images seem like a better quick test because we can see them at a glance. Maybe ask it to make a comic?
246.
by skybrian 2mo ago
It’s true that they’re destroying old books. The claim that they’re destroying rare books seems to be repeated without much in the way of evidence.
247.
by skybrian 2mo ago
You should cite your sources, and that might include an LLM. If you use AI chat as a search tool and then read whatever web pages it finds to check them for yourself (as you should), then you can cite that instead. Citing an LLM is a sign
248.
by skybrian 2mo ago
In many places you couldn't even buy "starter houses" like they had anymore. They wouldn't meet modern building codes. When people do manage to buy a house, it will likely be a nicer one.
249.
by skybrian 2mo ago
That kind of spending is included under "retirement spending" in the article. They estimate that substantial amounts go to heirs after that, but mostly among the affluent.
250.
by skybrian 2mo ago
I agree that the output is more likely to be valuable when it’s shorter than the input. And when they use tools to do research and try things, it can be a lot of input that comes from all sorts of places.
251.
by skybrian 2mo ago
The coding agent I use (Shelley on exe.dev) has a pulldown menu where you choose the LLM when starting a new conversation. There is no lock-in. Conversations are not precious. Although, I do use ChatGPT more nowadays because their subscript
252.
by skybrian 3mo ago
I don’t see this as a big deal in practice. Conversations contain a bunch of junk anyway, so removing it from the context is usually good. In my repo, I have a notes directory. I ask the AI to write a markdown file with what it learned, wha
253.
by skybrian 3mo ago
On the other hand, I suppose if the OS on a TV stick ran in a hardware-enforced sandbox that restricted network access to certain necessary domains, it couldn't be used for scraping websites and ad fraud? It's not being sold as a
254.
by skybrian 3mo ago
Even if we had an answer to the halting problem, from a practical point of view, it's solving the wrong problem anyway. There's no point in distinguishing a function that would run forever from one that would exit after a century.
255.
by skybrian 3mo ago
The FCC tests electronics for radio interference. Perhaps they could test electronics for Internet behavior like this too? Some manufacturers will try to cheat on the tests, but we have AI security checking now, so maybe that would make it
256.
by skybrian 3mo ago
Attempting to prevent websites from learning about their audience is difficult. For example, Lego has a "play zone" and it has other pages geared towards adults. You get an immediate popup asking if you want to go to the play zone
257.
by skybrian 3mo ago
They are being trained to try lots of unlikely alternatives and to be persistent. This often works well when searching for security bugs or counterexamples to famous math conjectures. But maybe it doesn't work so well when caution is r
258.
by skybrian 3mo ago
Vibe-coding something ambitious fits pretty well with all three. (Assuming a different kind of hubris.)
259.
by skybrian 3mo ago
> Below is an example of Kimi K3 using this for Unsplash, a free images provider that models commonly use to find hero visuals and other filler images. On top is the chain-of-thought reasoning, and below is the Unsplash CDN image that th
260.
by skybrian 3mo ago
Maybe the AI doesn’t need access to it by default? It could be hidden behind another tool call.
261.
by skybrian 3mo ago
Maybe it would be good if more electricity generation became available for other purposes like heat pumps and electric cars? Although, perhaps it doesn’t help as much if it’s in the wrong place. Also, cheaper RAM would be nice.
262.
by skybrian 3mo ago
It seems like there could be a filter so that the AI can only see the text when it’s clear that a user could read it, and it’s okay if the AI misses some text. This might involve actually rendering it, though.
263.
by skybrian 3mo ago
Why is it possible to have hidden text in a Word document? Why should the AI have access to that text?
264.
by skybrian 3mo ago
The hypothetical is that everyone will eventually agree with the author that AI is not useful. I don't see what could possibly change everyone's minds at this point. People have too much direct experience with it. At best, it migh
265.
by skybrian 3mo ago
You can do whatever research you like, but when you attempt to apply it to predicting the future, you're still speculating. Past performance is no guarantee of future results.
266.
by skybrian 3mo ago
Can't you link to whatever you want in your own blog post?
267.
by skybrian 3mo ago
Maybe this is more interesting in the context of the HuggingFace incident? Suppose you have an AI and some access controls for what it can do. If the access control is too broad or has a bug, can you still detect suspicious activity and slo
268.
by skybrian 3mo ago
There are some dubious metaphors in use here. Public health is not about math problems to be “solved.” “A solution exists” is just a milestone. “We just have to use them” - yes, but leave out “just,” there is more to do and it’s hard. The a
269.
by skybrian 3mo ago
Here's an example making the rounds, originally from 2022: there is systemic bias in Americans' understanding of US national statistics about all kinds of subgroups: https://yougov.com/en-us/articles/4155
270.
by skybrian 3mo ago
This is about saving money by using the right tool for the job. If you have a system that does a lot of mundane, repetitive work, you don't need a frontier model to do it. It doesn't mean frontier models aren't good at harder

More ›