Warren Koch

The specific comment you were disparaging was saying it is a dumb strategic move by leftists to reject AI technology. "on the same order as the green parties ditching nuclear to support burning more coal" is apt. Couldn't agree more. Lefist positions on AI are pathetic right now

Open source plays a backstop role, and is doing so just fine so far. We have many options, and they will become more relevant in time. Personally I forsee distributed consumer RL training as very viable ever since the INTELLECT-2 paper and expect we'll have our own compute pool in time

You're going way too far on purity testing and missing the forest for the trees here. Everyone new to a field starts with easy corporate tools and moves to open options as they get more sophisticated and informed. Their behavior is fine.

Instilling interest and nuanced discussion on AI and how to improve public options is fine. Calling people bootlickers is not. Take your aggression somewhere else

It is just as insidious and idiotic to enforce anti-AI stances by you shouting down normal AI enjoyers with moral grandstanding.

The absolutely key detail is that people NEED to be using these tools and keeping up if society is to ever stand a chance. They can not be relegated to the right

I think going around saying "people's brains are actively being captured" because they said learning how to use AI is important is a strange stance to open with, especially when we clarify it's open source AI.

Took a while to try and out-left each other but I think we're on the same page now

Parent companies should never be trusted regardless of their stated structure. If it doesn't live in immutable code on an immutable blockchain, it ain't free enough yet (and even that can be improved)

As someone who has thoroughly mapped out how to scrape all social network data and mass exodus social networks with a precommit-based rebellion (I'd call it Mutiny) - yes, it still irks me to see any company owning network abstractions. We should fix that soon too. User effort is the bottleneck

Open Source as an ideology and as a goal is not neoreactionary capitalist. It's a public option.

You can complain about it not going far enough yet (I do too), but not about the goal itself.

I am well aware of what is not "free" yet (and for that matter, what is not even "open source" like some of the AI models colloquially referred to as such).

My point is that the process of open source is a continual push towards "free", and it's one we're getting quite good at and will improve

I don't know what you're arguing for here - I guess that open source doesn't solve anything because it doesn't solve *everything*?

Or perhaps that we should somehow reverse AI technology itself (impossible) and how dare we be optimistic about open source AI?

Empty pointless whining

Either way, thanks to an open source project we have a guaranteed floor on how much a social network company can exploit, capped at just a bit over raw energy costs. We are fine with them acting as utility companies - nobody is complaining about water treatment companies and electric cos

BlueSky exploits that by paying people's hosting in hopes laziness will lead them to become the central network, but at any point if people decide they've abused their power they'll put in the effort (or ask their local AI) to self-host or cloud-host with a more trusted provider

The technical infrastructure is free, the hosting isn't. But it is incredibly cheap per user to host, so it's primarily just a matter of people not bothering to make the effort to do so. It's just effort - which AI will alleviate by doing it for you

Facebook is not free and not open source. You're literally on an open source project alternative to facebook/twitter called BlueSky, created for the sole purpose of undercutting their greedy underlying cost of siphoning user data.

This is the end result of software - we optimize til it's truly free

Except the red queen's race ends at "free". We are purposefully driving it there - because we are all trapped in the race as producers regardless, and need to at least have it benefit us as consumers.

Open source is literally the only path to mitigate this

Downstream is fine. Open source is a backstop against corporate overreach not a frontliner. We just need to ensure there's always a cheap public option that gives most of the value for free.

They should go for broad appeal - policy matching whatever the public overwhelmingly wants in polls, like universal healthcare.

Problem is their donors won't allow it. So instead its just socially-right tactics like chuggin' a beer every time someone says 'woke'

Okay, well enjoy basing your world view based on vibes and feelings rather than factual analysis

Sounds like remixing in your own personal experience. Will be quite interesting to see bots which make art factoring in theirs

Throw my posts in there too. It will give you all 'true's or 'partly true's. Though I'm *sure* it's all conspiracy propaganda, even when you ask it to cite sources

If you're really not including extrapolation and remixing of existing concepts as a valid new style, then you've already tossed out the entire history of art. Nothing is truly unique, it's all a melting pot. Though I would interested in seeing an AI given zero human data trained from scratch

At least hedge. At least make plans to treat these capabilities seriously and make sure they're not monopolized by corporations

At the very least, people should be treating these capabilities with respect and allocating appropriate risk to these improvements happening. (And as a senior programmer who has researched AI for years now, there really dont seem to be any scientific reasons it wont continue scaling)

Whether or not you believe AI will be adopted and demand will go up to make AI ventures profitable is not the point.

The point is the tech itself is certainly racing forwards in capability, and will be capable of having its own impacts on the world regardless of whether the general public adopts it

ChatGPT - AI Art Quality Debate Shared via ChatGPT

Not true and not how it works. It would derive its own styles from a corpus of just photographs of nature. It has already created its own styles which are distinct. It doesn't need new input data, what it needs is better organization of existing data and synthetic data

chatgpt.com/share/67becf...

As an open source AI advocate I say the only way to prevent this is getting the tools used by as many people as possible and to keep up with the innovations. If they can't ban AI the public tech will keep improving too, and hopefully be used for good. charities, communities, govs, safety nets

Trump's authoritarian security state crackdown is a great way for them to privatize all the gains from increased AI efficiency and prevent the general market from competing. That's all they need to win. (plus a ban on AI for everyone else)

If all goes well, they never will profit - but they will simultaneously erase the book value of most industries that used to have thick profit margins but are now forced to operate at razor thin ones from competition. I'm sure they're hoping to make a monopolization play to ensure they still win

While this will boost supply of everything by making it cheap to produce (both services and goods), it means far more competition and narrow profit margins. So it's not surprising or particularly notable that AI companies aren't profiting yet. They are creating a product that kills all markets