Upol Ehsan1/
OpenAI quietly shut down its "AI" detector.
Did shutting it down undo the harms?
No, its Algorithmic Imprint lives on.
Here's how ⤵️
#AI #ML #chatgpt #mastodon #responsibleAI #AIethics #generativeAI
2/
A) will it bring back the indignity faced by those who were falsely accused thanks to the detector's false positive rate? Unlikely. Think of all the students who perhaps got a failing grade thanks to the failure of the technology
B) will it undo the mistaken mental model such technologies instilled in people-- that AI could be used to detect if text is AI-generated? Unlikely.
3/
C) will it undo the mistrust the advent of AI-generated text sowed in our hearts? Unlikely.
As our work on the Algorithmic Imprint states, while algorithms are made of "soft"ware, they leave *hard* imprints on society. This is why we need to think twice before deploying algorithms because shutting them down does not undo the damage they do. Their Imprints live on.
//END
Raccoon🇺🇸🏳️🌈@upol
This is why greater awareness of AI and its impacts, with real knowledge and not misinformation, is so important, and why these Sam Altman types being allowed to hype their AI as if it is somehow infallible in all these public spheres is so damaging. I want governments to start regulating it, but I worry that the people causing the most problems are going to be the people with the most input on how those regulations work. I want serious research and writing to be done on impacts of these things, but I worry that the companies will drown that out.
Sometimes I worry that we've already reached the point where we are no longer in control of the AI, even without it acting autonomously, even without it having general or strong intelligence, simply because the AI is controlled by corporations, and they are controlled by profit margins, and not actual humans.
There is no one with a stake in the matter at the wheel.