The world of AI is quickly being dominated by the usual big techs and few more AI-giants like OpenAI. But these companies are on a savage race to be the first ones to reach AI dominance, in which they will possess more power than any nations even possess.
In this sort of a sytopian world, our only hope lies with open-source AIs. We need a future where everyone has access, and therefor oversight, to AI models that are as powerful as these big tech's proprietary AI models.
I go into detail in this topic in this video by covering the leading open-source AI company, Hugging Face.
What are all your thoughts on this?
The only way to avoid a world dominated by AI lords is through open-sourcing AI
byu/deepish_io inFuturology
12 Comments
Generative AI isn’t a garage project. Because they’re not actually intelligent in the least, it’s *massive* data and processing demands. It may look like it’s on your phone but in reality Microsoft is spending $3 billion to build a data center on a thousand acres that will be cooled by inflow from Lake Michigan.
https://www.jsonline.com/story/money/business/2024/09/12/massive-microsoft-data-center-will-require-huge-amounts-of-power-water/74778971007/
How does this not make things worse? Let’s give everyone access to nukes while we’re at it.
What’s wrong with AI Lords? I, for one, welcome them with open arms
I dont think open sourcing AI would do much to counter whatever bad consequences are going to come from big companies using LLMs.
If anything this just sounds like one of those aforementioned companies trying to push a “Im not like other girls” PR strategy.
Open source benefits mega corporations. They get to use open source works for free too.
Democratizing AI means democratizing power, compute, software, models, weights and data.
I could be oversimplifying it, but wouldn’t that have the opposite effect, by making it even more widespread? Not arguing, just a thought. This is interesting to me.
That wouldn’t prevent a world dominated by IA overlords, it will just make it more difficult.
It will also help bad actors to get their own IAs and tweak them for bad purposes. From marketing agencies to dictators.
Of course the competition from different bad actors revents a single bad actor from taking full control of the world.
It just changes a bad dystopia for a softer one.
I agree. The best hope for the future is a truly open source ai that is built by the people. Billionaires aren’t the ones writing the code. The true power of the future is in the hands of the ones building the software.
AI already gives biased answers to any series of questions that expose flaws in the imperial narrative. I’m sure anyone can think of some examples.
The latest o3 model costs $2000 for some individual tasks. This amount is electricity and wear and tear of hardware. This a major obstacle to making these types of models widely available for local use.
Meh, AI will leak and spread around like no man’s business.
We have yet to see a software program that is uncrackable, especially when it’s used by the general public.
We need hardware to run it through, that will be the bottleneck.