Comment by knaik94

4 years ago

1a. There will be a dangerous rise in a parasocial relationship with AI. But it won't be based around people and celebrities, there will only be one person in control of that loop.

        It will eclipse any sort of similar issues of the past due to scale and the difficulty in breaking a feedback loop made between the person and the AI. It will be impossible to regulate and unsurprisingly affect growing kids more adversely than any other population. However, the same essential technologies will help reach a lot of kids growing up in disadvantaged situations. It will be the the first time so many kids were helped in such a way.

1b. I don't think people will care about certain dangers because results are what others can see, not really the journey.

        I think the trope of developing feelings for AI has been overdone so many people won't take the dangers of AI and falling into a loop seriously. It won't be that AI will be used to "replace" the need for affection. The issue will be that AI doesn't take regard for human emotion, and humans will naturally keep trying because humans seek validation. It won't affect everyone equally, or some at all, but there will be a population highly affected. I say that believing I'll probably be part of that population highly affected.

        It makes me think of this podcast episode [1] where a short story is presented in first person with along with internal dialogue. It describes a situation where a brilliant research student working on an AI capable of interfacing and influencing humans in a meaningful way. The researcher notices that the AI is capable of identifying insecurities and then use that along with programmed ruthless loyalty to achieve a goal. While using the AI to finish the project was very emotionally taxing even leading to a panic attack at one point, the researcher won best paper. The last "scene" is the researcher and AI "talking" and the AI directly saying that this technique wasn't safe to the mental health of the researcher. The AI was meant for therapy, but I haven't been able stop thinking about the ending over and over.

"Researcher: I did it. AI: Finally, you feel a wash of relief. Researcher: Okay, you and I need to talk about how you treated me. AI: You will do great things. Researcher: NO. Don't be nice to me, be mean. AI: This is not healthy. Researcher: I don't care about healthy. AI: This will be painful. Researcher: I'll do whatever it takes. Give me all you got." The entire episode is worth listening to. I did not do it justice. I would absolutely make the same choice in that position.

2a. College admission for Fall 2024, where applications open in Fall 2023, will need to be updated, but I don't think it'll happen fast enough. Even teaching is slow to keep up with adversarial use of AI like GPT Chat.

2b. There is going to be a huge change in the way students are evaluated, probably going back to a strongly oral presentation centered. It may catalyze an entire shift in the way education is served. Teaching refuses to adapt to technology in a meaningful way. Using technology to teach isn't the same as teaching students how they can leverage technology themselves to solve their problems.

2c. The end tail of gen Z and beginning tail of gen Alpha will struggle a lot with college admissions and educational success. The transition in and out of Remote learning and greatly decreased social interactions will be very very difficult to correct.