Viewing profile — dumberquestions

I think you're underestimating both their reliability for standard problems and the usefulness of that level of reliability.

So this demo is around 90 times faster than typical speeds for the same model at openrouter, and around 30 times faster than the absolute fastest option available (Groq).

I wouldn't expect companies not sharing their weights today to be any more likely to share them if they're on hardware, this doesn't sufficiently hide weights from a local user.

Which model? Or how many active parameters?

Ride hailing apps promised to decrease traffic and social media apps promised to make people more social, in both cases the opposite happened and this is just off the top of my hea…

Robot-taxis don't offer the same convenience and status gain of a private car, while a smartphone does every single thing a landline can and more, this is not a good analogy.

I think we're underestimating how relatively recent these social phenomena are. Just a few decades ago, the average person had very limited exposure to people outside of their imme…

Have people's standards for what a house constitutes gone up, or are affordable options that people would be very willing to buy less available now?

I predict that teenagers with irresponsible parents will continue to use social media and online anonymity will get worse.

Well he's saying what a lot of people desperately want to believe.

Why would they decay in usefulness?

They're quite obviously Moroccan.

AI would be profitable today if all companies weren't constantly racing to train the next biggest model, that's the whole reason the numbers look whack. This is basically the first…

The dedicated research team is getting moved to other areas, the Alphafold system itself and its database will likely continue to get maintained, but this means there probably won'…

A lot easier actually, writing a plot from scratch is very different from adapting an existing one.

There are some reasons the story could be inaccurate in some ways: OAI stands to benefit if people think their models are strong, and they have a history of doing things with dubio…

Ironic that polymarkets were being advertised as helping society make better predictions.

If I'm going to read it wrong might as well have an LLM to blame.

You're right, should've gotten some LLM to summarize it instead of skimming.

"..and in some benchmarks like DeepSWE by Datacurve, we observe up to 65%, all at a lower cost per output token." "3.6 Flash delivers higher precision with fewer unwanted code edit…

Do you really think popular opinion always gets reflected on policy?

I wonder if certain checks should be put in place to prevent a few companies from hoarding a valuable and limited resource many other industries rely on.

Things are relative.

The trend might've stalled or even reversed if it weren't for AI, we can't just assume the same end was written in stone.

It's still not obvious that AI is uniquely environmentally destructive compared to some other "non-essential" industries, and it gets even less obvious when that's weighed against …