233mhz

Would you carry 100kg of very nice powerlifting weight plates in your car because they're very nice even though you never use them?

The entire world is based on trust and doing what you said you would do, everyone more or less got the memo except trump. The thing with trust is that it takes decades to build and one sentence to break

Which scholars? The ones hiding rampant child sexual abuse for decades? The ones asking for genocide of their neighbours or the ones calling for war on the degenerate western world ? Or perhaps the smaller ones?

I can't wait to hear about their super informed views and opinions about clankers.

And usage will grow accordingly, as usual.

My car engine is a miracle of efficiency compared to the stuff we had a hundred years ago, but overall cars pollute more than back then

> my house uses an average of 1300kWh per month.

And the average European house uses like 3000kwh.... per year. We're already wasting insane amount of energy, adding 2-10% more is definitely huge

This is a techie shit posting forum not a court of law, calm your tits down

If enough people keep saying it I'm sure they will nerf it

A lot of people don't see tech as a way to improve their life but some kind of fashion trend they have to keep up with. There also is the problem of people who were either bad devs or non dev at all who now get this omnipotent machine they can talk to in natural languages and it just does things for them, no matter how retarded the things are.

> My problem with plugins is that I don't want to install stuff from third parties.

Then ask the model of your choice to implement the plugin for you, I created 3 this morning just for myself

It sounds fancy and technical. Like when people use "orders of magnitude" six times per thread on this forum but no ever ever use it irl

"china bad because china copy"

Meanwhile there is an american backdoor in your bios, cpu and router

The whole thing around the model that handles system prompt, skills, agents, cache, etc. For example codex, claude code, pi, oh-my-pi, ...

Complexity has to play a role, google chrome has more LOC than windows XP... At this point everything is a pile of piles of shit built on top of a gigantic pile of shit.

Today I ran out of ram, the "teams" desktop app was taking 2 gigs just standing there doing nothing, you can run four whole windows Xp on the same amount of ram, comfortably. My inactive chatgpt chrome tab alone was taking 900mb

What's new is "smart" decision models than you can supposedly use on anything without additional training.

If you have a very narrow use case you can train a BERT based decision model on a laptop an hour if you have good data to train it on. It'll answer faster than the roundtrip to clef/jev and use <1gb memory

Benchmarks are only part of the story, most workloads don't match what you see on these.

For example sonnet is supposedly the same price as opus from medium and up, I tested it on a real world task at my day job and sonnet was a good 30-50% cheaper

The problem is that you need to find an equivalent or worse psycho who don't mind chopping heads as a 9 to 5

If you spent even a single day in these you'd know it's much more inhuman than death

Now try again with local and short term impacts on the whole ecosystem instead of saying "Ackchyually water never leave the atmosphere so it's unlimited"

> something happened once in the past and turned ok hence you should never ever worry about other vaguely similar things ever again

That's how it sounds to me

> If they really had a much more powerful model, it would make absolutely no sense to sit on it.

Makes total sense if they don't have the compute and can't serve it in an economically viable way. Also it lets you build things no one else in the world can build as fast as you until it's released

> We already know OpenAI has "bel" that is MUCH better than astra and is being used internally

I mean, isn't it almost a guarantee that what we get is a gimped version of what they use internally? They probably already serve themselves next gen level models at 1k+ tps from cerebras machines hosted on perm while we get quantized astra/opus at 50tps on a good day