@matthewswspence@matthewstoller Like, they surely didn't prompt "find some vulnerabilities" on the whole codebase, right? They probably had some agentic system to break it up into smaller tasks, which, if they're trying to show off their new model, would be as liberal with token usage as needed.
@matthewswspence@matthewstoller Isolating parts of the code doesn't necessarily seem like cheating to me, depending on how it was done. I'm no expert but I assume Mythos had some tooling around it which did exactly that (isolate individual parts of the code) to search for vulnerabilities.
@airkatakana What makes software projects "good" is intense attention to detail, security, speed, etc. That last 1% is what separates products people will actually use from slop.
@markvalorian@TheIllegit Not really. The topic "Ice" doesn't even correlate with the increased interest from the shooting. The search term does (although has a pretty high floor preceding that, likely from unrelated searches).
@Kyochi_Myogo@TheIllegit It was a comparison of a topic and a search term, to get a sense of which is more accurate.
But yeah, that is a better topic for this particular issue. And actually if you compare that with "Renee good" and "Ice", it becomes clear that "Ice" is nonsense.
@Kyochi_Myogo@TheIllegit This is what the 30 day graph looks like for both search term and topic. The search term actually much more closely follows my perception of public interest in ICE (shooting happened on the 7th, ICE search term has a strong peak on the 8th)
@Kyochi_Myogo@TheIllegit I think it suggests that google isn't properly aggregating searches for the topic "Ice". In comparing topics, you are relying on googles aggregation of related search terms. Either way, its certainly not an exact science.
@TheIllegit If you compare the search term and topic for each, altogether, you get this for average interest over the course of the week. ICE, search term, is the most popular in the vast majority of states. Even when compared against the *proper* topic "Chicago Bears"
@TheIllegit I assumed you did based on the fact that your original post labeled red as "ICE" not "Ice". "ICE" is not a valid topic on google trends. I wouldn't exactly trust how google is aggregating related search terms for a topic its not even properly capitalizing!
@TheIllegit If you compare both as "search terms", the graph looks quite a bit different, and, regardless, if you look at interest over the course of the week, searches for ICE are clearly much more popular.
@danbri@LewisCTech I’m pretty sure he’s “skeptical” that AI can replace coders at a large scale, not that it can, in a pinch, splurt out a solution to a problem that’s been solved 1000s of times over.