nathansherburn

As someone in the "not set for life" club, I worry a lot about my peers who are blind to the reality of AI progress. You don't have to believe AI is entirely hype to be anti-AI. It can be bad AND real.

I learnt a new word today. Thanks!

I've actually been following this project for a long time and it's none of the above. They're simply testing what a set of frontier models can do when given a goal and left to their own devices.

I agree this outcome is very painful to see and I really feel for Rob. It's clear people (myself included) are completely at breaking point with AI slop.

In this specific case though it's worth spending 30sec to read the website of AI model village to understand the experiment before claiming this was sent by Anthropic or assigning malicious intent.

Wouldn't frequent reauth be beneficial for stolen sessions?

E.g. If you set your session timeouts to a ~1 day then by the time your session cookies are up for sale on the dark web, they will be expired.

The article doesn't mention this and it's the main reason I advocate for auth sessions that are as short as practical.

How does divorcing Europe make the US stronger against China?

With who in control? A Russia, China and US alliance? This seems very unlikely to me. All signs indicate we're moving to a multipolar world as far as I can tell.

China owns an increasing majority of global GDP and Trump seems to be either taking a path of peace, populism and or stupidity by pulling America out of its global alliances.

A great way to get around this is with an edge function from deno deploy.

I think this is spot on. The confusion for me comes from the fact that, as far as I can tell, I've never met a prig in real life. And yet they seem to be the biggest political issue of our time. Is it because I live in Australia and it's more of a US thing? Or is it because I'm not online as much maybe? I find it really confusing.

Not sure if relevant but I thought this looked very cool.

https://www.squiggle-language.com/docs

I don't think the purpose of the checkbox was to collect useful data but an any case storing it as a date would probably solve the issue.

I'm still using the built in MacOS terminal. What am I missing? I mostly do webdev - maybe terminal features are more important for other types of programming?

How do you think reasoning happens in our brains? I wonder if it's more like an LLM than we realise?

Open source software is one of our best and most passionately loved inventions. It'd be much easier to have a nuanced discussion about "open weights" but I don't think that's in Facebook's interest.

Thanks for the write up!

I just wanted to chime in to say the Foliate code is really beautifully written too.

I was able to hack together an epub audiobook reader by running a custom version of foliatejs in Microsoft Edge (which has an amazing and free text to speech engine called "read aloud"). It's VERT hacky but you can try it here: https://nathansherburn.github.io/foliate-js/

Works on mobile and desktop Edge.

Reports are that Trump is okay.

Politically I disagree with so much of what Trump says and does but I would never wish this on him. We can't let the extremists on either side win.

AI will always be far better at languages than I'll ever be and I expect it to get much better very quickly. But I'm still learning my partner's language and don't think I'll stop any time soon. I think it's interesting and fun in and of itself. It's also a great scaffold for learning about another culture and learning to respect and understand folks from different walks of life.

I agree we want to democratize AI and we should be very, very weary of powerful people trying to get a monopoly on AI.

But I'm not ready to say absolutely everything should immediately be shared with everyone. At least not until it's clear we know what we're dealing with.

I know it can be hard to trust people but the fact is we have to. Even today, there are many people with the power to end the world (nuclear weapons, viruses etc) but we trust the people who have these capabilities not to abuse their power. We do this because we don't want anyone to have the right to launch nuclear weapons. And I think that's wise.

I definitely don't know for sure but AI may be another one of these technologies.

Either way, you can fight against regulatory capture, the downsides of capitalism etc without being an open source absolutist.

I share your concerns and think you're broadly correct. I think it's worth adding some nuance though.

When you drill into specifics there are almost always exceptions. For instance, in your example about sharing research, there are certainly some types of research we shouldn't automatically and immediately make publicly available like biological superviruses or software vulnerabilities.

I think the same can be said about AI. We should aim to be as open as we can but I'd be hesitant about being an open source absolutist.

Is there any evidence this actually works? My understanding was that "prison as a deterrence" was largely ineffective.

Just three hours ago I switched back to Linux after a few years on MacOS. The only thing missing was the amazing text copy tool I was using, "Rex" [1]. What a coincidence to see this post on the front page a few hours later!

Side note, what a breath of fresh air Gnome on Fedora is!

[1] https://github.com/amebalabs/TRex

There is research out there that shows use of white space can improve things like reading speed and comprehension.

An example for text paragraphs: https://www.semanticscholar.org/paper/Reading-Online-Text%3A...

I think it really depends on individuals though. If you can memorise a dense screen of buttons you'll be able to work faster, avoid scrolling etc. But it'll make the UI harder to use for people who don't use it regularly.

Ultimately, every UI has to strike a balance. If you do it right you'll piss off both sides equally.

Just want to say thank you for your enormous effort on this. I'm extremely picky about software but Firefox on Android is one of the best pieces of software I've ever used. And I use it every day - probably more than any other app. I still remember how happy I was when youn folks got those adblocker extensions working on mobile. Thank you.

That's true. There are many other viruses that we don't publish for good reasons though.

Fair point. I think the thrust of the argument still stands. Open source is generally a fantastic principle but it has its limits. I.e. we probably shouldn't open source bomb designs or superviruses.

Any examples of the web component code they spit out?

Thank you! I can't believe I never thought to do this!

I'm surprised given the makeup of the hackernews crowd there aren't more people who appreciate this here.

I only know a few folks who work at the big AI labs but it's very clear to me that they are personally worried about existential risk.

Do people here not have friends and family working at these labs? I just figured people here would be more exposed to folks working in the leading labs.

Audiobookstore.com are fantastic. I've enjoyed so many of their audiobooks over the years.

I suspect it's your add-ons. I've been using Firefox mobile for years and I don't think I've ever had a site that worked better in Chrome.

Caveats: Sometimes launching apps from the browser only works in Chrome and some demos of bleeding edge experimental Chrome features obviously don't work in Firefox.

Am I correct that this only shows GUI apps that can be installed via Homebrew? I'm sure I have some command line utilities installed that aren't showing up in Applite.