Skip to playerSkip to main content
  • 10 hours ago
AI models seem friendly, patient, and able to adapt to their users. Researchers warn that this may make real encounters with others more challenging and result in emotional consequences.
Transcript
00:00AI tends to flatter its users. It will agree with their opinions and confirm
00:04their preferences in order to create a feeling of closeness and affirmation.
00:10When we get used to those kinds of interactions it will change what it's
00:15like for us to go back out into the real world and to experience human beings who
00:21do not interact with us in this sycophantic way. And that could make us
00:25more emotionally vulnerable to real human interactions, could make us more
00:29impatient. It could make us just want to give up on being able to and wanting to
00:36interact with other human beings which is very dangerous.
00:40Humans can be difficult. Arguments with AI language models on the other hand are
00:46almost non-existent. Manufacturers are interested in users spending as much
00:51time as possible speaking to the chatbots. The way that the LLM has been designed in
00:56this sycophantic way that it wants to continue the interaction, the discussion
01:01for as long as it possibly can, we are being fooled into believing that there's
01:06something more going on behind it.
01:08This can lead to a loss of social skills. Early studies suggest that spending more time with
01:15chatty PT is linked to increased feelings of loneliness.
01:20Can we use AI to help with loneliness? It fuels this narrative that AI is a magic bullet,
01:28is a band-aid, yeah, that it's going to ultimately solve this problem. And the moment we can recognize
01:35that it's not the thing that's going to solve all of our problems, then we will start to look at
01:41it as
01:41one piece of the puzzle. And I think that's how we need to move forward with AI in general, but
01:47in particular AI in care context with this companionship, loneliness epidemic questions that we have on our mind.
01:55Around the world, this technology is being introduced everywhere. Developers profit from humans testing AI tools and algorithms in daily
02:03life, generating data that can be fed back into their AI's.
02:07This also results in many products being released while still in the development phase.
02:14We're all members, a part of this experiment across society. We never gave our consent to it. We didn't click
02:22on the terms and conditions of this big experiment.
Comments

Recommended