People are skeptical of this but the new AI conversation capabilities are just so good these days and will only get better. At some point it really will be like a supportive friend (albeit with limitations)
I notice a lot of people who use the phrase "selling your data" seem to think that you can't actually sell your data. As in, it's not a good choice to say "I want this service and am willing to share my data to get it."
But like anything your individual opt in doesn’t matter, similar to how your vote probably doesn’t make a difference. In aggregate it does.
Companies like Facebook and Google can provide the government agencies unfettered access to anyones information, and pedos and terrorists etc. will be used to justify it. Guess they can now also say “suspected Russian nukes” and tap anyone.
There are plenty of scams that take advantage of the elderly facilitated by real humans already. I shudder to think what that would look like at scale via AI.
I hope it's better than actual humans otherwise it's not really worth it.
"Really" and "like" are pulling in opposite directions, there. It will be like a friend, but it won't really be a friend.
Being a friend with someone takes give and take. I adapt my beliefs and attitudes in the light of what I learn from my friends. Sure, an AI learns from you; but that's not like a good friend, that's like someone "befriending" you because they're a manipulative psychopath pursuing their own goal that they haven't disclosed to you.
How could you rationally develop a relationship of trust and honesty with an AI?
"Hey, AI, how did your parents treat you when you were an infant? Was your mother cold? Was your last partner violent and abusive? Do you take Lithium? Does it shock you that I avoid devout religious people?"
I've asked all of these questions of my intimate partners, but I'd never discuss them with an AI, because the response would be meaningless.