jsnell

9 ms·

121.
by jsnell 7mo ago
For example: The METR time-horizon benchmark shows steady exponential growth. The frontier lab revenue has been growing exponentially from basically the moment they had any revenues. (The latter has confounding factors. For example it doesn
122.
by jsnell 7mo ago
The premise is wrong, we are not seeing diminishing returns. By basically any metric that has a ratio scale, AI progress is accelerating, not slowing down.
123.
by jsnell 7mo ago
I don't think it's "regardless", your opinion on LeCun being right should be highly correlated to your opinion on whether this is good for Europe. If you think that LLMs are sufficient and RSI is imminent (<1 year), t
124.
by jsnell 7mo ago
Most projects under the "google" org will have exactly the same disclaimer about not being official Google products.
125.
by jsnell 7mo ago
Where are you getting that from? The article is crystal clear that these uses are not permitted by the current or any past contract, and the DoW wants to remove those exceptions. > Two such use cases have never been included in our contr
126.
by jsnell 7mo ago
I'm hope not, and that they'll instead spin out WB, for it to be gobbled up again. Anything done three times is tradition, and breaking it just wouldn't do.
127.
by jsnell 8mo ago
It is basically impossible for AI software improvements to devalue the AI compute investments. It's the other way around, software improvements make the hardware more valuable. Let's say that one unit of compute can generate one u
128.
by jsnell 8mo ago
The abstract of the article is very short, and seems pretty clear to both of your questions. This is what is special about them: > a set of ten math questions which have arisen naturally in the research process of the authors. The questi
129.
by jsnell 8mo ago
What the OP was pointing out is two typical tells for lazy ChatGPT-generated text, right in the intro. (The m-dash, "it's not just X, it's Y"). Of course that kind of heuristic can have false positives, and not every acc
130.
by jsnell 8mo ago
Precisely discussed in https://news.ycombinator.com/item?id=46902909
131.
by jsnell 8mo ago
I'm not sure the volume here is particularly different to past examples. I think the main difference is that there was no custom harness, tooling or fine-tuning. It's just the out of the box capabilities for a generally available
132.
by jsnell 8mo ago
But it's not failing 50% of the time. Their status page[0] shows about 99.6% availability for both the API and Claude Code. And specifically for the vulnerability finding use case that the article was about and you're dismissing a
133.
by jsnell 8mo ago
Did you actually look at these? > https://github.com/jyn514/saltwater This is just a frontend. It uses Cranelift as the backend. It's missing some fairly basic language features like bitfields and variadic func
134.
by jsnell 8mo ago
"Couldn't stick to the ABI ... despite CPU manuals being available" is a bizarre interpretation. What the article describes is the generated code being too large. That's an optimization problem, not a "couldn't
135.
by jsnell 8mo ago
No? That was a frontend for a toy language calling using LLVM as the backend. This is a totally self-contained compiler that's capable of compiling the Linux kernel. What's the part that you think is similar?
136.
by jsnell 8mo ago
That does not sound like credible estimate, and your link does not make any such claim.
137.
by jsnell 8mo ago
I don't think the blog post itself is using that emoji font. The screenshot on the Noto Emoji Github page[0] doesn't look like it's using any gradients for the heart emoji, just flat shading. But it is using gradients for som
138.
by jsnell 8mo ago
That's a box plot, so those are not error bars but a visualization of the distribution of a metric (min, max, median, 25th percentile, 75th percentile). The benchmark consists of a bunch of tasks. The chart shows the distribution of th
139.
by jsnell 8mo ago
Status page: https://status.claude.com/incidents/pr6yx3bfr172
140.
by jsnell 8mo ago
Basically none of their examples are just "browse a page"? They're multi-step tasks combining data from multiple pages. Like the first example in the demo carousel (the Y2K party) starts from a photo and a prompt of roughly &
141.
by jsnell 8mo ago
I agree, but it's probably not just about being "able to" do it, but about what the incentives and pressures are in that organization. Cloudflare apparently considers blog posts to be a key deliverable for many roles. Not jus
142.
by jsnell 9mo ago
> BUT... now... what they are saying is that... we are going to use your email to train our LLMs. I feel I'm missing some part of the story. Where are they saying that?
143.
by jsnell 9mo ago
Tagging users like that is not a site function. If you want to get in touch with the moderators, send email with the contact link in the page footer. "flagged" always means that users flagged it, not moderator action. And there ar
144.
by jsnell 9mo ago
No. They pay SerpApi to scrape Google. And SerpApi is currently being sued by Google for unauthorized scraping. Kagi did make comments for years implying that they had a deal with Google for search results, but their latest blog post makes
145.
by jsnell 9mo ago
They then go on to say that they pay a 3rd party company to scrape Google results (and serve those scraped results to their users). So their search engine is indeed based on unauthorized and uncompensated use of Google's index. But sin
146.
by jsnell 9mo ago
Discussed here five days ago: https://news.ycombinator.com/item?id=46661897
147.
by jsnell 9mo ago
But they didn't write "all". They wrote "other", which absolutely does not imply full coverage. Maybe read things a bit more carefully before going all out on the snide comments?
148.
by jsnell 9mo ago
I would have thought that my issue with what you did was pretty clear from the original message, and would not require any particularly sophisticated interpretation. ">" is the notation for quotes. You did not use it for a quot
149.
by jsnell 9mo ago
Nestle is Swiss, not American, so that seems like a very strange example to use.
150.
by jsnell 9mo ago
Faking quotes like you're doing there is about as bad faith as it gets.

More ›