People are already complaining about how the AI training data from recent forums are “contaminated” with outputs from other AIs, if you want something “purely human” to work from then historical pre-2023 data is the best bet.
In the final analysis, nobody cares what Harold Q. Dumpington bought from Amazon in the week of June 4, 2017. That information is technically still stored in Amazon’s databases, but (1) Amazon already has access to it, so encryption is a sort of non-issue, and (2) nobody cares.
The reality is: socially engineering a password or setting up a “man in the middle” attack in a coffee shop WiFi is a hell of a lot easier than attacking encrypted data, but even those attacks are relatively rare, and usually executed against corporations with money. As tempting as it would be for some hacker to get into Jennifer Lawrence’s e-mail or Chris Pratt’s Amazon purchase history, it seems that it’s really not worth the effort to anybody, except in some edge cases.
Putting aside the whole question of what people might want to feed into an AI, why would anybody want that data AT ALL?
You’re making claims about what everyone (and everything) to ever live after this point in time is going to care about. That’s unfounded and kind of presumptuous.
If an AI was being trained to “be” a specific person, why wouldn’t their history of Amazon purchases be useful as part of building up that persona? Or on a broader scale, wouldn’t patterns of purchases be useful for modelling cultural patterns?
People are already complaining about how the AI training data from recent forums are “contaminated” with outputs from other AIs, if you want something “purely human” to work from then historical pre-2023 data is the best bet.
In the final analysis, nobody cares what Harold Q. Dumpington bought from Amazon in the week of June 4, 2017. That information is technically still stored in Amazon’s databases, but (1) Amazon already has access to it, so encryption is a sort of non-issue, and (2) nobody cares.
The reality is: socially engineering a password or setting up a “man in the middle” attack in a coffee shop WiFi is a hell of a lot easier than attacking encrypted data, but even those attacks are relatively rare, and usually executed against corporations with money. As tempting as it would be for some hacker to get into Jennifer Lawrence’s e-mail or Chris Pratt’s Amazon purchase history, it seems that it’s really not worth the effort to anybody, except in some edge cases.
Putting aside the whole question of what people might want to feed into an AI, why would anybody want that data AT ALL?
MC Frontalot has a song about this, Secrets from the Future.
You’re making claims about what everyone (and everything) to ever live after this point in time is going to care about. That’s unfounded and kind of presumptuous.
If an AI was being trained to “be” a specific person, why wouldn’t their history of Amazon purchases be useful as part of building up that persona? Or on a broader scale, wouldn’t patterns of purchases be useful for modelling cultural patterns?