this is a high bar, but fable 5.2 still lacks that high percentile human trait of dementedly caring about the work they are doing. you can tell it to read an entire archive, it will just get the general gist, and just assume that that is ok. importantly, not a capability problem.
35
the thing London has over basically anywhere is the co-existence of all of the different eras of western history. if you don't get inspired by 60's brutalism, maybe 30's art deco will do it. if not, maybe medieval/regency/georgian/baroque. all directly beside each other.
1
62
Prague is one of the most beautiful places on the planet, from the second floor up
WARNING: In Prague you’ll see lots of shops that seem to be selling weed. AVOID!! They are fraudulent tourist traps. At best, you’ll get CBD products sprayed with unknown chemicals, at worst you’ll overnight in A&E - as has happened to a number of visitors lately. Stay safe! ❤️🤍
1
3
476
struggling to see why steven pinker's views matter, this isn't a man with power of the meaningful kind
1
60
Found this pretty lazy in the MIT tech review. And another reason why I’m a bit put off by any AI discourse that focuses so much on any specific word as a mark of progress/threat, even if in this case I ended up agreeing with his larger points around creativity. This article tries to show that LLMs don’t ‘reason’. Observe:
1. Vague appeals to authority i.e. ‘a scientist’. What is a scientist?
2. If you get to decide what words mean, i.e. decide what counts as shortcomings to a definition of reasoning, then of course your argument is going to make sense
this applies to a lot of the ‘serious thinking’ that ends up in places like this and Wired
3. But if your definition of reasoning doesn’t count some of the stuff models and agents are currently doing right in your face, then that definition just cannot be that good.
1
74
LLMs continue to struggle to contribute meaningful intellectual work with vague scope. e.g. give Opus 5.5 a task to integrate a lot of material into some cohesive analysis on a topic, or what research/truth it should push you towards, and it will give everything exactly equal weighting. When for a skilled human, the point of that process is cutting out the inputs that turn out to be not as useful. Forgetting is just as important as learning when it comes to cutting through noise on a topic.
1
2
44
they should also integrate a giant claw like the ones at those arcade machines that picks people up at random and drops them in different spots
or not at random, but guided by guests secretly voting with buttons under the table
Maybe this is a stupid question but why don't dinner parties put everyone's chair on a conveyor belt so they can talk to different people throughout the evening?
1
82
I think blogs and essays will quickly move in the direction of more interactive types of media, where many more things can happen within, and it feels more like a world the reader interacts with and moves through. This is a very cool genre of thing that it has never been so easy for individuals to produce.
35
also be very wary of free xg data as a consequence, risk of it being more misleading than your eyes.
98
if you've put down 22k with a mainstream bookmaker on big ticket events, and you're only $17 down, you are either very lucky, or strangely good at gambling which means you might have a skill you should think about putting to use
2
416
symbols, simulacra, cognitive shortcuts, identity, desire, Baudrillard, Lacan
enjoy
alreadyhappened.xyz/p/she-sa…
1
4
183
Here’s the overall problem: if you disincentivise winning gamblers out of most bookmakers, you accidentally make that bookmaker more powerful in their ability to play their customers. You end up with gambling being an even more unfair game than where we started.
1
83


























