this post was submitted on 18 Jan 2024
15 points (89.5% liked)

Solarpunk

5492 readers
1 users here now

The space to discuss Solarpunk itself and Solarpunk related stuff that doesn't fit elsewhere.

What is Solarpunk?

Join our chat: Movim or XMPP client.

founded 2 years ago
MODERATORS
 

Hi all, I'm working on a Solarpunk world building project and I want to know your thoughts on one of the main features of my world. To preface all this and provide some context, my world is an alternate-history with a divergence point sometime in the 2020s. The divergence was caused by a vocal and technically-skilled group of Green-Anarchists that labelled themselves as "Dawn".

Dawn did a whole host of things to ween people off of Capitalism and into my Solarpunk world, I've gone into immense detail on this but I doubt it's relevancy to my question so I'll omit all those details, but there was a tipping-point in which Capitalism crumbled and gave into Dawn's Anarcho-Solar world.

To make sure the world stayed Solarpunk and to give people stress-free lifestyles, they developed 1 AI and 1 AGI. The AGI manages all Dawn technology, such as Dawn power generation, carbon-capture, a global hyper-loop etc and the AI makes sure no one tampers with the AGI (For those unaware, AGI is Artificial General Intelligence, so for example Skynet is an AGI since it can think and do many things, but ChatGPT is an AI because it can only do text).

Most people in my world wont ever have to think about the AI and AGI, it is taught in my education system to make people aware in case of catastrophe but it mostly manages itself and is monitored by the longest-serving Dawn members.

I simply want to know if machines like this can exist in Solarpunk with it remaining Solarpunk, and if people like the idea or not. If you want to know more about my world building then feel free to ask! Thanks for your time in advance :)

P.S. I should mention that AI and AGI are mirrored across 8 different instances and for the most part work independently of each other, meeting only when strictly necessary. This is to give even more defense against tampering and error.

Edit 1: Changed title from Overlord to Background, Overlord implies oppression which the system doesn't do.

top 41 comments
sorted by: hot top controversial new old
[–] keepthepace 8 points 10 months ago* (last edited 10 months ago) (1 children)

I simply want to know if machines like this can exist in Solarpunk with it remaining Solarpunk, and if people like the idea or not.

As someone who went into robotics and AI 20 years ago because I strongly believe they make a whole new range of utopias possible, I'll of course answer YES! Of course! And there are not enough fictions like that. Most sci-fi is dystopian and 90% of solarpunk dreams of a low tech future where everyone is a farmer which, to me, is another dystopia.

People here already mentioned Banks' Culture series, that's the only example of such things. I wish we had more, and if you ever need technical help on the AI design side, I'll be happy to answer questions.

Even though this is not my personal opinion, you are in good company by using the definitions you use for AI and AGI. Technically you are talking about language models and multimodal models. It remains to be seen if multimodality is required for general intelligence but that's certainly a plausible hypothesis for a sci-fi setting.

We are in dire lack of high-tech solarpunk around here, more power to you!

[–] Penguin_Rocket@lemmy.ml 1 points 10 months ago (1 children)

We are in dire lack of high-tech solarpunk around here, more power to you!

It sounded a bit counter-intuitive at the beginning, but I thought a bit about your point of view and I finally agree with you. We still need to have a bit of high-tech in solarpunk and low-tech. Of course, we need to drastically think how to use it ; everybody having a last-generation smartphone with more than enough power to send people on the Moon just to stay entertained all day long is a nonsense. However, AI and data science can be really beneficial especially for scientific purposes.

[–] keepthepace 3 points 10 months ago

It is not the tech that is problematic, it is the culture around it, the economic system and the incitations around it. Having enough power to run AIs locally would allow everyone to carry a personal doctor and lawyer and has tons of applications. The fact that it allows such an intense level of communication across people all over the world is something that we take for granted but that has actually been world changing.

We don't want high tech for the only purpose of having high tech, but we want it for the things it brings: knowledge, communication, automation (in the pursuit of labor abolition). If we could find a low tech way to achieve these, I would be all for it, but for now, high-tech is the enabler of these things.

[–] poVoq 8 points 10 months ago (2 children)

For a Solarpunk story it is a very weak plot element as it basically just handwavy says the AI is solving all the hard problems. This is both unrealistic and also more akin to transhumanist narratives that claim that some imaginary and not yet invented technology will solve all of humanities problems.

[–] keepthepace 3 points 10 months ago (1 children)

Why do you assume it is handwavy? AI will solve many problems and create many others. That's the staple of sci-fi to explore these.

Suddenly workers are not necessary anymore => How does that change our urban organisation?

Wealth inequalities can't really be used as a tool of oppression anymore => What oppression device remain and how do we get rid of it?

more akin to transhumanist narratives that claim that some imaginary and not yet invented technology will solve all of humanities problems.

Not "all", but what type of utopian sci-fi does not rely on technology to solve some of our current problems?

[–] poVoq 2 points 10 months ago (1 children)

Its handwavy because it appears to be the central plot element. Current "AI" is far from AGI, and even assuming some one comes up with an AGI in the near future, this would not solve many of the world's problems.

The utopian sci-fi that does not rely on magical tech is called Solarpunk 😅

[–] keepthepace 3 points 10 months ago (1 children)

AGI will solve many problems. I mean, if you are talking about a 50 years apart future, it is far more realistic to try and imagine one where AGIs exist and solve many of the things they were designed for than to imagine they won't appear or won't change anything.

And I don't see what in OP's post make you think they will make their AGI work on "magic" or if they make the assumptions it will solve "every" problem.

For heaven sake, Asimov describes such futures in the 50s! And they are neither transhumanist nor utopian. He explores the impact of intensive robotics on human interaction and social structures.

How can solarpunk call itself a sci-fi movement if it refuses to go beyond today's tech? or sometimes even just refuses to acknowledge current tech?

[–] poVoq 1 points 10 months ago* (last edited 10 months ago) (1 children)

There is no AGI and current LLMs are unlikely to become such. Maybe, my predictive powers will turn out to be wrong, but I think there will not be an AGI in 50 years either.

Good Sci-fi is mostly a social commentary on the present. Sure, sometimes some magical technology is a useful rethorical device and sometimes placing the story into the far future can be used to great effect to drive home a point...

Yes, there is hard-sf that sometimes builds an interesting story extrapolated from existing technology, but AGI is not something that can be used in hard-sf as we have no realistic clue how to make an AGI.

Edit: we have no shortage of human level intelligence on this planet, yet somehow this doesn't seem to solve most problems 😅

[–] keepthepace 2 points 10 months ago* (last edited 10 months ago) (1 children)

You don't need an AGI to automate 99% of the labor. No one in the field thinks we can remain 50 years without getting an AGI.

And saying that a sci-fi author should refrain on imagining the consequence of AGI is... an interesting opinion to say the least. The whole field is about imagining how future tech will impact us!

we have no realistic clue how to make an AGI.

We do. There are many possibilities, many leads. From multi-modality to just scaling up, experts in the field rule none.

[–] poVoq 2 points 10 months ago (1 children)

I did not say so, but it wouldn't be hard-sf but rather speculative fantasy about some hypothetical technology.

I think a lot of people "in the field" will be up for a lot of disappointment in the coming 50 years, and not because human intelligence is so amazing 😅

[–] keepthepace 2 points 10 months ago (1 children)

speculative fantasy about some hypothetical technology.

So every story that include worker robots, spaceships, time machines, fusion power, moon bases, submarine bases, are not sci-fi? Dune, Foundation, The Culture, the Three-Body Problem, Blindsight, hell, half of Jules Verne books, are not sci-fi then?

Sorry, we are not talking about the same thing then.

[–] poVoq 1 points 10 months ago* (last edited 10 months ago)

They are not hard-sf is what I said. And most of the technical examples you mentioned are not comparable to AGI.

[–] UmbraTemporis@lemmy.dbzer0.com 1 points 10 months ago (1 children)

It doesn't "solve problems" per say, but informs people of those issues before they become damaging. Since most technology has some kind of Dawn-imprint on it it can interface with almost everything. Humans still have to go out and perform maintenance etc.

All the AGI can do is optimize existing Dawn technology and adjust the allocation of power to each purpose.

[–] poVoq 8 points 10 months ago (1 children)

That wasn't my point.

You plot is basically: "underdog group out of nowhere invents incredibly advanced cybernetic AGI way beyond what is currently technically feasible and by doing that saves the day".

This as a plot is neither very realistic nor shows a feasible way how we might solve today's issues.

Global coordination issues are the hardest problems, and while some sort of AI might play a role in this in the future, I don't think taking that as a central plot element in a solarpunk story is a good idea.

Maybe you could center the story around every day people trying to fill in the coordination gaps that an imperfect but hailed as a saviour AGI leaves wide open, but I guess that would come closer to a cyberpunk story then.

[–] UmbraTemporis@lemmy.dbzer0.com 2 points 10 months ago (1 children)

I hear you, the narrative I have planned takes place around 50 years after the shift to Green-Anarchism, so I'd imagine if I continued like that but made the AGI a new invention it'd be more believable, with Dawn having more time to develop Sci-Fi technology? However I am in two minds to shift the narrative back to the emergence of Dawn and the transition to their world. Doing that would make it much easier to include more action into the story and give me more limitations which helps me massively in worldbuilding, it was personal preference to set it in the future since I love Sci-Fi :D.

[–] keepthepace 3 points 10 months ago (1 children)

Don't worry, a lot of us want to hear about the society post-AGI.

[–] UmbraTemporis@lemmy.dbzer0.com 1 points 10 months ago (1 children)

I've made the descision to keep it in but I've lowered it's scope quite a bit and made it an experimental recent creation in the timeline. This way it wouldn't make sense for my society to totally depend on it, like they originally did, but I can still explore a world where people rejoice in AGIs existence instead of man vs machine.

[–] keepthepace 2 points 10 months ago (1 children)

Aw! Don't let the naysayers here sway you too much. The dependency is certainly a worthy consideration to have (as we discuss here over our totally dependable internet infrastructure) but don't make it miss the fundamental society changes that it can cause!

[–] UmbraTemporis@lemmy.dbzer0.com 2 points 10 months ago

Oh yeah absolutely, I'll include it in such a way that people are beginiing to substitute the tougher jobs for it while still making sure people are trained and ready to take over at a moments notice.

[–] Marsupial@quokk.au 5 points 10 months ago* (last edited 10 months ago) (1 children)

Check out Ian M. Banks The Culture series. AI and humans do-exist, AI are also super nice and provide stuff for us because we’re basically the equivalent of puppies compared to their level of superiority.

Oh and each of the AI that are ships have names like “Funny, It Worked Last Time...”, “A Series Of Unlikely Explanations”, or “Very Little Gravitas Indeed”.

But in my opinion yes, but we have to co-exist making democratic decisions together. SolarPunk is anarchy, so you can’t just have a machine overlord lording over people.

[–] UmbraTemporis@lemmy.dbzer0.com 3 points 10 months ago

This is what I intend, the AGI will do some monotonous tasks autonomously but will be in contact with Humans almost constantly and will give them final say. It mostly monitors and alerts if any fundamental Dawn technology is acting up and also adjusts the allocation of power globally, determining how much goes to each settlement, and giving the excess to itself or to "climate reclamation" technology.

[–] wanderingmeomeo 4 points 10 months ago* (last edited 10 months ago)

I'm not really sure, or, well, at least AI in what people tend to associate with nowadays. I think a solarpunk future is a future where degrowth is a main ethos for the development of the society. AI, as it currently exists, relies on global scale exploitation of not just artists, writers, but also people in the Global South living in abject conditions and are forced to accept mentally and physically damaging jobs to help "AI" to exist in the first place. Using AI, as far as I know, also comes with huge environmental costs simply because of how energy intensive it is to run and train its huge dataset. So the existence of stuffs like ChatGPT or Midjourney is a no go, unless "human ingenuity" could do something like being so cost effective energy wise that technology like that would require way less energy to run, or somehow the colonization of the universe would happen, but for the latter to happen will neccesarily require the kind of economy that centers growth above all else like capitalism.

Imo, Solarpunk technology should be local and open source and easily understandable for communities and individual to use, the factor of being environmentally friendly is also very significant and I don't think AI infrustructure would guarantee to fullfill those requirements.

That's the reason I would never understand the obsession with nuclear power or AI techinology of some solarpunk enthusiasts. These are all infrastrcutures that requires a degree of centralization of human and resouces, which drastically constrasts what solarpunk should be about. A solarpunk future is not fully automated luxury communism, but a low tech communism that still guarantees the well being of all. Low tech does not exist simply because of accessiblity, but also the capital E Environment.

Furthermore, the basic issue of human creativity being ripped off is also a thing that should be addressed. I don't belive in intellectual property, but we shouldn't need IP to protect arists from exploitation without consents. A solarpunk world is a world that respect human autonomy, so if an artist do not consent for their works to be used in training AI, their wish will be respectfully followed through. As an artist in art communities, I don't think that many people are fond of submitting their works to an AI so that others could code a drawing, so the dataset required for AI would not be enough, thus render AI useless. That's just creatives though, maybe scientists and researchers would hold different views? If that's the case, circle back to the question of energy intensity inherent in AI technology.

That's just AI in a very specific sense though, I don't know what constitute an AI. If AIs are just machines that can beat human at chess, or to predict patterns of local weather, or regulate irrugation system for farming, or organize documents, then sure, there's no reason why it should not exist in a solarpunk future. But the existence of such machinery should be put under scrutiny. If an AI could not be operated without significant environment impact, it has to go. Impressive techonology is not just codes, it's also very real, very material hardware that possess impressive processing power, and with it, impressive amount of energy, resources, and human power.

To address your AI specificly: I think your AI is HUGE, and I'm not confident about its place in a solarpunk world. Your AI would exist in some versions of communism, but solarpunk? I don"t think so.

[–] Five 3 points 10 months ago (2 children)

I simply want to know if machines like this can exist in Solarpunk with it remaining Solarpunk

I think this question is a great prompt for an actual Solarpunk story.

Forget AGI, even creating 'AI' seems to require human exploitation on a global scale. All current Large Language Models' (LLM) existence relies on denying intellectual 'property' norms and erasing the authorship of millions of small human authors, artists, and creators, while those same norms are enforced brutally against them in the interest of rich creators and corporations.

I know anarchists take many stances on the use of LLM technology, but I don't think anyone denies these technologies are powerful and world-changing. Like not driving a car or living a strictly vegan lifestyle, some anarchists have measured the harm with the benefit, and decided to accept that boycotting LLM technology is an ethical stance they can't individually afford.

But what would an anti-capitalist 'AI' look like? Is it possible to build an alternative to the LAION Dataset that respects the human authorship upon which it depends? How would such a system survive against its capitalist competitors? How would tools built on the dataset differ from the ones currently available? How would abuse of the system be prevented or mitigated? How would the world react to an anarchist LLM?

I think answering these questions could be the basis of a solarpunk version of the cypherpunk Cryptonomicon. It would be the kind of well-written story I would like to read.

[–] schmorpel 5 points 10 months ago (1 children)

I think the problem is not with intellectual property but with artists being forced to create art for profit. Can we decouple art from commerce completely and set up society in such way that we can afford to create art for pleasure? If I needed to work only two days a week to provide for myself I could create art in the remaining time. With art not being a commercial product anymore, intellectual property wouldn't be an issue, plus people could create art from the depth of their hearts and not geared towards commercial success.

[–] Five 5 points 10 months ago* (last edited 10 months ago) (1 children)

Forcing people to perform a skill they would do for free both drains the joy out of creation and undermines cooperation between creators. I agree that being forced to create art for profit is part of the problem.

One of the foundational coups of capitalism was framing respect for artists as a form of 'property' - twisting the interests of culture creators and landlords into the same emerging political system. Now that capitalists believe we're at the 'end of history' this fiction is becoming less important.

When someone creates beauty, people will want to show gratitude. Even in a post-scarcity world, there is incentive for jerks to misrepresent the origins of their work and to plagiarize. I think people will create art regardless of their political or economic situation, but I would like to think that whatever replaces capitalism would be a system where artists flourish because of their environment and not in spite of it. Intellectual property is old bathwater, but attribution and respect is our baby.

[–] schmorpel 3 points 10 months ago (1 children)

The age of content. When I grew up, millions of teen artsy folk like me were told we could 'earn a living with our creativity' in marketing and adjacent areas aka 'sell your soul to capitalism'.

[–] Five 1 points 10 months ago

Yeah, you nailed it.

[–] federatingIsTooHard@lemmy.world 3 points 10 months ago (1 children)

it's long past time for so-called "intellectual property" to be abolished.

[–] poVoq 5 points 10 months ago (1 children)

Agreed, but that doesn't make it right for multi billion dollar companies to take artists work without their consent and turn it into a commercial AI.

At the very least those AIs would need to be released to the public domain if they were trained on a public body of works.

[–] federatingIsTooHard@lemmy.world 1 points 10 months ago (1 children)

that would come with the abolition of intellectual property

[–] poVoq 4 points 10 months ago (2 children)

No, because AI runs in data centers and only offer an API.

[–] keepthepace 3 points 10 months ago (1 children)

Mistral-7B and Mixtral are 2 top-scoring LLMs that you can run on a personal computer. Admittedly a beefy one for the latter model.

Stable Diffusion generates very good image on personal computers.

All three are part of the hundreds of very good open source AI models that the industry is struggling to keep up with. OpenAI and MidJourney are the only one to manage to keep up the pace. Think about it: there are billions poured monthly into the industry and a bunch of researchers and hobbyists beat that on home PCs. Even mfing Google does not manage to keep up!

All these models can be run locally with no third-party between your prompt and the result.

And if you dont have a good GPU, you can join the AI Horde, that provides share computation to everyone.

90% of the non-tech news about AI talks about OpenAI but when you follow the tech news of the scene, the FOSS (Free Open Source Software) community is actually winning a very important battle without the general public being aware of the battles going on.

[–] poVoq 3 points 10 months ago (1 children)

I was specifically referring to commercial AI, and besides that: those ones you mention are under quite restrictive licensing and can't really be considered in the public domain either.

[–] keepthepace 1 points 10 months ago (1 children)

Apache 2.0 for the first 2 ones. It is essentially the same as public domain.

CreativeML Open RAIL-M for Stable Diffusion.

The Linux kernel is under more restrictive licenses.

[–] poVoq 1 points 10 months ago (1 children)

I stand corrected on the first two, but Stable Diffusion's license is significantly more restrictive than the GPLv2.

[–] keepthepace 2 points 10 months ago

I'm myself no fan of OpenRail, but I think it is more of a result of FOSS ceding to silly fearmongerers than a trojan to close down models.

[–] federatingIsTooHard@lemmy.world 2 points 10 months ago

with no intellectual property, the code can just be exfiltrated

[–] averyminya@beehaw.org 1 points 10 months ago (1 children)

If we can get low powered machines (seems like analog computers built for the specific task are a decent bet, among things like AI RAM modules) running it on an independent solar grid isn't infeasible. Our immediately available technology doesn't have this but it can exist with just a little growth.

For real world inspiration of the technology check out Vertisatium's video on Mythical AI doing recognition in real time on analog chips, along with Intel's announcements on their AI RAM modules. Should give you some good framework.

[–] keepthepace 2 points 10 months ago (1 children)

Our immediately available technology doesn’t have this but it can exist with just a little growth.

We know how to generate MWs out of solar power and some models run on raspberry pis. Our immediately available tech has this. I run a very decent LLM (Mistral-7B) on my machine and it must draw 500W max, that's extremely doable.

[–] averyminya@beehaw.org 1 points 10 months ago (1 children)

500w is what my PC w/ a 3080 runs at with stable diffusion, with everything including monitors and etc it usually caps at about 700w - while the analog equivalent IIRC was about 80w max under load (I'll have to double check that though). Powering the peripherals like monitors and speakers draws more than sifting through an LLM with that tech! And that's a more overarching issue when you scale it up per-home, worse if per individual.

I don't necessarily think it's infeasible, it's more about applications and purposes, like why use a 500w dedicated power supply if the same purpose could be scaled down leaving 420w for other tasks. Similarly, we use cloud clusters for stable diffusion, mid journey, dall-e and such all aren't running on one computer alone but a bunch of computers connecting to a server to process the loads in queue. So while the PC's aren't working on "the same task" they are all working on the same goal. For a community server like a stable diffusion discord this work is spread across volunteer PC's that are (mostly) already being used for day to day tasks anyway. But even then, say just 5 people volunteer their PC's that's now 2,500w all just to speed up a queue. Whereas a specified instrument cut down to 80w ran by 5 people is only 400w.

Basically what I'm getting at is right now almost regardless of what technology we have available, we are brute forcing AI. Using CPU and GPU tasks to accomplish something that can be done with an analog computer in a quarter of the time with 98% accuracy. For former is just not sustainable by comparison.

On one hand, making breakthroughs towards making current technology better and feasible is a good thing. On the other hand, why not use all available options and utilize the benefits of each. Consumers being able to use AI locally is awesome, but consumers using an energy efficient option would be even better.

I came to this conclusion with the solarpunk ethos in mind, and again it's not that I think it's infeasible or shouldn't be done or anything like that, just that we have a bit of tunnel vision with digital computing and it would be interesting to see what could be accomplished with a wider scope. Especially since it is somewhat just a breadboard with copper wires ;) lol

[–] keepthepace 2 points 10 months ago

Agreed, there are certainly more optimal ways to do. The applications of LLMs are being made right now, and multimodal robotics systems are coming out of the lab this year. My point is that we should not believe that they pose any kind of energy waste problem if they do a useful function.