dash2

18 ms·

361.
by dash2 9mo ago
> Why do you think we have been fighting the 'war on cancer' since the 60s? Actually, we have made huge progress in the war on cancer, so this example doesn’t seem to support your narrative.
362.
by dash2 9mo ago
Specifically as someone in the UK, where doctors are free but extremely hard to get hold of, this is quite interesting to me.
363.
by dash2 9mo ago
These points seem to be arguments against giving your health data to anybody, not just to an AI company.
364.
by dash2 9mo ago
Just wanted to say these are great comments – you are good at explaining complex economic concepts clearly.
365.
by dash2 9mo ago
Polymarket isn't on the other side of those bets though. It's a two-sided market. I don't think Polymarket even sets the terms of the bets, or adjudicates them?
366.
by dash2 9mo ago
Europe can make alternatives to US tech Then why are there approximately no European tech companies? You remember that FT graph....
367.
by dash2 9mo ago
How readable is the code produced by opus, when you try to interact with it manually?
368.
by dash2 9mo ago
I left desktop Linux in 2010 because everything did not just work. Looking at the responses to your comment, it seems that basic stuff like wifi still doesn't always just work. If it's been true since 2010, I think the problem is
369.
by dash2 10mo ago
> Who knows if it was better in the nationalised days I do. I was there. It wasn't.
370.
by dash2 10mo ago
Could you give some examples of the Western propaganda machine producing xenophobia and racism?
371.
by dash2 10mo ago
People downvoting you may think that this is an uninteresting quibble: we may not find it very surprising that sometimes Waymo asks for human guidance, and we don't necessarily think "autonomous" is an all or nothing designat
372.
by dash2 10mo ago
It's good and gave me some queasy chuckles. But it's not as fun as the "HN Front Page" from a few days back. I suspect there's kind of a feedback loop here. An algorithm does X well; it becomes easy to churn out X;
373.
by dash2 10mo ago
I would be interested to hear HN views on Richard Hanania's take on Epstein: https://www.richardhanania.com/p/the-inverted-reality-of-eps... .
374.
by dash2 10mo ago
Even just setting aside the criminal/salacious side... it is just so interesting to see the connections between the elite. Steve Bannon photographed chilling with Noam Chomsky, such a surprise. God, and Robert Trivers is in there too,
375.
by dash2 10mo ago
> it often makes early assumptions and fails to validate them, which can waste a lot of time Is this baked into how the models are built? A model outputs a bunch of tokens, then reads them back and treats them as the existing "state
376.
by dash2 10mo ago
I find it helpful to, let's call it "vibe-check" the headline statements in articles like this. Essentially I'm using the llm as a glorified search engine, and also hoping it will have, not zero bias, but less bias than
377.
Are you vibe-coding an open source project?
2 points by dash2 10mo ago|0 comments
378.
by dash2 10mo ago
C'mon. Five years ago I could still tell myself Siri is fine. Now it's embarrassing.
379.
by dash2 10mo ago
> It supported networking and peripherals. Yeah, like a little stylus that you could use to touch the screen and move the mouse pointer around your mini Windows desktop! There was even a way to right-click. Seriously. Cue that sweet pict
380.
by dash2 10mo ago
I tried the new vibe-coded Mandelbrot viewer and it didn't seem to work well on Safari. I could only get it to zoom in once, and most of the keys didn't work. Maybe the author hasn't done enough manual testing?
381.
by dash2 10mo ago
Why "click, scroll and drop", rather than "click, point and drop"? The latter seems more general, works in 2D, and doesn't require two-finger mouse scrolling.
382.
by dash2 10mo ago
Yeah, they seem to be moving from innovators to exploiting their user base. In particular, I interpret their AI failures as the innovator's dilemma. They have loads of highly paid people designing and creating beautiful UIs. A chatbot
383.
by dash2 10mo ago
They made the most popular RPG last year already - why do you think it'll take 20 years for them to make the most popular shooter? For that matter, the Singapore-HQed SEA makes Free Fire, which topped Google Play in 2019.
384.
by dash2 10mo ago
You have to understand that while the rest of the world has moved on from 2020, academics are still living there. There are many strong leftists, many of whom are deeply censorious; there are many more timeservers and cowards, who are terri
385.
by dash2 10mo ago
This would be true if all training were based on sentence completion. But training involving RLHF and RLAIF is increasingly important, isn't it?
386.
by dash2 10mo ago
I've heard of loads of people who would prefer British rule over Chinese rule. The reasons are obvious: free press, free speech, fair trials. These things have gone far, far backwards under China.
387.
by dash2 10mo ago
More obviously, the objects in the mirror aren't actually reversed!
388.
by dash2 10mo ago
No freedoms are ever unlimited, so that's not an interesting point. The question is, were they useful? A lot of HKers seem to think so.
389.
by dash2 10mo ago
The important bit, which they don't talk about, is that custom instructions seem to have been used to trick ChatGPT into giving dangerous shell commands as help.
390.
by dash2 10mo ago
I think the name is not good. It sounds like "stool app". Among other things, "stool" means poo.

More ›