Read the article, a lot of the misidentified lawmakers were black
Not surprising that the corporate slop machine is also racist.
It didn’t “misidentify” anyone, it did what it was designed to do and now they know which parameters they need to “adjust” and will now identify those that would oppose them.
It’s a feature, not a bug!
What scares me about AI (in this case more LLMs) is not necessarily how they’re weaponized today but how they centralize power structures.
People are gradually becoming accustomed to their corporate flavor of LLM as their single source of truth. That is tremendously centralizing force for control.
unfortunately i think we passed that a while ago when we siloed off the internet into social media sites.
Counterpoint, small decentralized LLMs that can run on local machines are empowering people to do more things and away from big tech. My non-tech friend switched to cachyOS thanks to ollama.
thanks ollama
I’d agree that it’s all a tool and it’s in how you use it, look at Reddit vs Piefed but just like this example it really feels like we’re losing the numbers game. I’d attribute that to friction of use.
It’s ok. We don’t have marketing money. We are always going to be the minority. You can only save whoever you can.
Great point. I think their original intent was total value capture of the emerging tech, but this ended up being a fantastic side benefit that they have no problem leveraging in a post-truth world.
