__alexs

5 ms·

31.
by __alexs 3mo ago
You are simultaneously arguing for a more complex and nuanced testing approach that demands much higher quantities of data as a result, and also against RCTs, which perhaps rightly you've identified as having suffered from the same kin
32.
by __alexs 3mo ago
The problem is that people are not listening to each other. Part of this is just working together is hard, but a bigger part is that there is status associated with being "in charge" of things as founder. The desire to feel like y
33.
by __alexs 3mo ago
Good stress test for your SQL parser I guess.
34.
by __alexs 3mo ago
100% this. I read these caveats in new models and all I hear is "we made sure this model has no idea about computer security." Such a weird thing to brag about.
35.
by __alexs 3mo ago
I find it hard to trust a ranking system that gives Sonnet a higher capability score than Fable.
36.
by __alexs 4mo ago
If the EU develops a market leading model to even control in the first place I think that would at least be a neutrally good outcome.
37.
by __alexs 4mo ago
Actually it's the opposite. Per mm of silicon it's massively less efficient and making enough chips and powering them is a major bottleneck right now. Worse, scaling to larger models requires more of our absolute best quality sili
38.
by __alexs 4mo ago
Oh sure the EU is terrible at all technology policy but it also seems like this is sort of a don't interrupt your enemy when they are making as mistake situation. If the US commit to handicapping their AI industry like this it's g
39.
by __alexs 4mo ago
I don't understand how this doesn't entirely screw the TAM for these companies? As an EU company I think I now basically have to consider US AI as hostile and avoid it.
40.
by __alexs 4mo ago
There are legal forms of killing. It is only via the application of our legal institutions that criminality is decided. No specific act on its own is criminal, there is no Platonic Crime.
41.
by __alexs 4mo ago
This doesn't illustrate anything. OJ was innocent until the second trial found him guilty, they just had a different variety of punishments available.
42.
by __alexs 4mo ago
Only our legal institutions and the frameworks they create can decide if any specific act is a crime.
43.
by __alexs 4mo ago
It goes to space like all the other heat does otherwise we'd be living on a ball of gently rotating lava.
44.
by __alexs 4mo ago
Have you read Don Quixote?
45.
by __alexs 4mo ago
What happened to innocent until proven guilty?
46.
by __alexs 4mo ago
To be truly impartial I don't think you should even evaluate their claims directly. This allows them to focus the comparison on things they care about, not the things you care about. Instead you should decide what problems you need sol
47.
by __alexs 4mo ago
A large part of vulnerability analysis is just having the time to crunch through enough possibilities. Expertise and smarts definitely speed this up but there's a lot of just turning the crank until something falls out. Even a relative
48.
by __alexs 4mo ago
The majority of people that drink espresso drink it with milk.
49.
by __alexs 4mo ago
This is private equity 101 no?
50.
by __alexs 4mo ago
There's no ethical consumption of... ad supported content.
51.
by __alexs 4mo ago
Surely this just uses state funds to train a pipeline of people to be immediately poached by frontier labs?
52.
by __alexs 4mo ago
I like Cursor as a product but the add on cost of $0.25/M tokens is just too expensive on top of the models.
53.
by __alexs 4mo ago
> Hans Boehm built a Java library for this in the 1980s and 90s Hans is such a prolific programmer that he wrote a Java library before Java was even invented?
54.
by __alexs 4mo ago
The crunchiness of the software rendering from that era is embedded in my memory like visual dial up handshake noise.
55.
by __alexs 4mo ago
I mean that kind of seems like exactly what's happening for AI to me.
56.
by __alexs 4mo ago
I'm not nearly that sick.
57.
by __alexs 4mo ago
Asked it to review some of my own blood test results and it immediately turned itself off and went back to Opus. Pretty disappointing.
58.
by __alexs 4mo ago
It's always worth understanding the consequences of non compliance before you'd decide to comply (and vice versa.)
59.
by __alexs 4mo ago
The quality of AI doomerism takes is matched only by the quality of AI boosterism takes. Ed's kind of interesting as a temperature sensor but I don't feel like you can really take anything he writes seriously.
60.
by __alexs 4mo ago
Human systems are not good at rapidly adapting to change. AI could be absolutely perfect and we'd still struggle to deploy it in a value generating way simply because it will exceed our ability to adapt. So tokenmaxxing might be the wr

More ›