yeah I'm also getting this for standard dev work, anything with kubernetes etc
completely nerfs the model because you can't let it do stuff over a few hours unattended because 90% it's going to switch to opus in first 10 minutes anyway
so seems best thing now is to have it write plans and then default to using opus for work anyway?
A lot of the soft and hard sciences use hacky matplotlib code to produce results and visualisation, without being necessarily data science
From the bits I've seen, I'd take claude-generated code any time over that written by maths, physics, biology, linguistics people. Even though I've seen Claude make some super-big mistakes while doing data analysis I'd guess it's already more reliable than most academics trying to code.
sure that's why we look at a mix of formal benchmarks, one longer analysis of a side-by-side, and various other people who we trust to form an opinion, all covered in the article - not intended to be a formal benchmark, there are enough of those.
yeah I don't think that's going to work - it would be kind of like "we're releasing model answers to all assignments but please only use them as a teaching aid and don't copy from them"
best to
a) adapt assignments so that agents are bad at producing solutions
b) have more scenarios where students have to do things in controlled environments. Universities managed to adapt to 'any solution you need is readily available online' so I don't think it will be that different to have several times a month/year where students have to go into a room with nothing but pencil and paper to prove what knowledge they have vs what they have the skills to access
It's annoying when people post slop with no substance, but it's somehow worse when there's actual substance _hidden_ by the slop.
I'd probably much rather have read the prompt for this article than whatever this is.
Anyway, I use my `buzzoff.wtf` slop site for this kind of thing. Originally built so I could figure out what startups actually did behind all their buzzword landing pages, but now also to get a summary of articles like this to decide if it's worth reading or not
> “Only two things are important right now in life: fertility and deep learning,” the University of Pennsylvania economist Jesús Fernández-Villaverde said at the conclusion of a recent lecture. “Everything else is noise. Once you start thinking about these, it’s hard to start thinking about anything else.”
Not sure if I agree but that's quite a memorable quote
so yes, it was clearly drafted by Claude. But I stand by the ideas, which I gave Claude to write this. We have a mix of automated and manual articles on the site, and the automated ones are all labeled Claude, and some people have been enjoying them. Happy to remove if against HN guidelines to have any claude-text.
How much software do you need and how many computers are there to run it on?
After combine harvester, we produced the same food with less people.
At the moment, it seems like hardware is the constraint. Companies don't have access to enough machines or tokens to keep all their devs occupied, so they let some go. Maybe that changes, maybe we already have too much software?
Personally I think we already had too much software before LLMs and even without them many devs would have found themselves jobless as startups selling to startups selling to startups failed and we realized (again) that food, shelter, security, education etc are 'real' industries, but software isn't one if it's not actively helping one of those.
I don't care how much maths and encryption you use, you can't get out of the fact that things can be anonymous (no one can know how you voted) or verifiable (people can prove that you only voted once) but not both.
- Switzerland usually gets around this by knowing where everyone lives and mailing them a piece of paper 'something you have'
- South Africa gets around this by putting ink on your fingernail
I've read quite a bit about the e-voting systems in Switzerland and USA and I just don't see how they thread the needle. At some point, you have to give someone access to a database and they can change that database.
Until we all have government-issued public keys or something, there isn't a technical solution to this? (Genuinely curious if I'm wrong here)
I haven't tried the $200 plans by I have Claude and Codex $20 and I feel like I get a lot more out of Codex before hitting the limits. My tracker certainly shows higher tokens for Codex. I've seen others say the same.
interesting we have a very similar internal flow - we like working in markdown but our customers want to leave feedback in Google docs, so we also have an md -> gdoc tool. We don't do the reverse as we ask them to only leave comments/suggested changes and we apply those directly to the markdown and re-export.
I ran into similar issues as you for the image handling, and the work around I use is to use pandoc to convert to docx as a first step and then import that as a Google Doc using the API, as Google Docs seems to handle docx much better than markdown from what I've seen.
This is a cool story but I'm really confused by the details. Like he seems to fly around and do pieces of this at a time, but then there's the weird bit of him walking 3000 miles in the US to get to the embassy, though that wasn't part of his 'walk'?
have you tried? I'm a South African living in Europe and visas are a nightmare.
Many europeans have never had to apply for a real visa in their life (I don't mean the online ones, or the apply on arrival ones, I mean the ones where you submit a 20 page form of personal details and hotel bookings and letters from friends you'll be staying with and bank statements and a full travel history) and they assume that I'm just making life difficult for myself by not doing some simpler option that they assume must exist.
I don't know about what visa options UK citizens have for the EU since brexit, but I'd be surprised it was as simple as "I feel like spending more than the 90 days I get".
