Ashtear

@Ashtear@piefed.social

• ⁨Comment⁩ on ⁨Visa says AI will start shopping and paying for you in 2026⁩ ⁨⁨2⁩ ⁨weeks⁩ ago⁩:

  This is where I sit on it. I think there’s a way to create a personal shopping assistant that would save people time, but the problem is I don’t make enough money to have a comfortable enough margin for the error rate it would certainly have, to speak nothing of whether it would find the prices I’d be satisfied with. I don’t know how many people are in affordable enough living situations where using this to save time would even be responsible.

  And that’s assuming the agent is working in the best interests of the user, which we all know isn’t how this is going to work. In general, I don’t know if I’ll ever be comfortable with agentic AI spending my money without it being codified into law that AI agents must have a provable fiduciary duty to end users. As far as I know, no one’s even talking about that.

• ⁨Comment⁩ on ⁨US adults social media usage by platform⁩ ⁨⁨1⁩ ⁨month⁩ ago⁩:

  Not sure how you got to that conclusion. The survey prompt was “Please indicate whether or not you ever use the following websites or apps.” 29% responded that they do not use the platform, less than 1% did not respond. That’s 70%.

• ⁨Comment⁩ on ⁨One in four unconcerned by sexual deepfakes created without consent, survey finds⁩ ⁨⁨1⁩ ⁨month⁩ ago⁩:

  That’s why I said local models. They aren’t automatically taking outputs and training updates on them.

  And yeah, we’ve all already had our likenesses folded in somewhere. That’s the bigger problem here.

• ⁨Comment⁩ on ⁨One in four unconcerned by sexual deepfakes created without consent, survey finds⁩ ⁨⁨1⁩ ⁨month⁩ ago⁩:

  Yes, if it was on a locally-hosted generative model, I wouldn’t be bothered if someone did this in my likeness. That wouldn’t be meaningfully different than using Photoshop to fake it ten years ago.

  Passing it around to their friends and gods know whom else is still just as reprehensible though.