I wouldn't be quite so pessimistic. We may have saturated the current approach, but I think there's a lot still left in terms of compression, attention, active parameters, caching etc. etc.

Did it ever occur to you that politicians signing a 'binding agreement' may not be the same as public interest? That they are often paid to hide the full implications from the public? Do you understand that some people actually want to live and raise family in a livable environment?

I understand that Ayn Rand fans who do not have to suffer the consequences of that philosophy or 'futurists' who think purely in 'make AI go brrr is cool' do not understand this, but some people place value on other things in life besides this and yes, they do deserve a place at the table.

As generally a tech optimist, but also someone deeply aware of the awful trajectory modern tech is on, such as their waste, war on general purpose computing, ownership, privacy and repairability, I think a lot of this needs rethinking and perhaps of brute forcing your way to stripping people of their livehoods by polluting, straining the grid and depriving people of places they'd want to live in, slow down and consider if there's not a smarter, more efficient, more humane way to move forward.

Failing to do so may produce a backslash that will harm even the things we love about our space.

The 'contract' is not sacred above everything. Especially when that contract affects a lot more people than those who signed it as 'binding'.

That being said, didn't know there's a RISC V effort as well now, so TIL.

Granted, this can probably be tuned for.

What do you think the end result of making regions unstable such as by unprovoked attacks on Iran is?

> you think

Well, maybe I am presuming too much there.

If someone enters my house uninvited, I don't care what they have to say to justify themselves because they don't have a right to be there in the first place.

Also, Israel designates tents dual use.

Designating an AWS location very likely used for targeting assistance during this war on Iran is a very different thing.

Let's not forget that as of now most models are like game cartridges i.e. they need a 'host' to run i.e. llama.cpp - they're not executable code by themselves.

I could see an argument for a security review prior to integrating a model into a highly sensitive industry/agency, but that is standard for any software.

It's certainly not an argument for banning open models because US AI corps don't want to compete with them.

By supposing, you can try to justify taking just about anything away.

> The only thing a ban would do is protect the American market from further downward price pressure on inference, protecting VC investors in the short term. But thats also an admittance that the American labs can’t compete on merit anymore

That's the argument. To be precise the publicly stated argument is that they're attacking American providers by distilling. The real aim is to eliminate competition because otherwise Anthropic and OpenAI are non viable and the US views them as crucial for winning the 'AI race' which they see as putting whoever wins it on top in terms of warfare/economic power etc.

The issue seems to be the US only likes competition when it is winning. Markets in Asia are meant for cheap labor and resources, they're not meant to actually compete. /s

Have you seen the massive blacklist Chromium maintains?