I'm unwilling to accept the discussion and conclusions of the paper because of the framing of how LLMs work.
> socio-emotional capabilities of autonomous agents
The paper fails to note that these 'capabilities' are illusory. They are a product of how the behaviors of LLMs "hack" our brains and exploit the hundreds of thousands of years of evolution of our equipment as a social species. https://jenson.org/timmy/
To the point of the paper, it has been a somewhat disturbing experience to see otherwise affable superiors in the workplace "prompt" their employees in ways that are obviously downstream of their (very frequent) LLM usage.
One very new behavior is the dismissal of someone's writing as the work of AI.
It's sadly become quite common on internet forums to suppose that some post or comment was written by AI. It's probably true in some cases, but people should ask themselves how the cost/benefit to calling it out looks.
They are then asked if they agree or disagree with a (presumably hypothetical?) company's proposal to reduce employees' welfare, such as replacing a meal with a shake. Two groups showed a different preference.
This makes me think about that old question of whether you thank LLM or not. That is treating LLMs more like humans, so if what this paper found holds, maybe that'd nudge our brain subtly toward dehumanizing other real humans!? That's so counter intuitive...
I can't tell if I'm just getting old, but the last 2 major tech cycles (cryptocurrency and AI) have both seemed like net negatives for society. I wonder if this is how my parents felt about the internet back in the 90s.
Interestingly, both technologies also supercharge scams - one by providing a way to cash out with minimal risk, the other by making convincing human interaction easier to fake.
12 comments
[ 2.6 ms ] story [ 35.0 ms ] thread> socio-emotional capabilities of autonomous agents
The paper fails to note that these 'capabilities' are illusory. They are a product of how the behaviors of LLMs "hack" our brains and exploit the hundreds of thousands of years of evolution of our equipment as a social species. https://jenson.org/timmy/
It's sadly become quite common on internet forums to suppose that some post or comment was written by AI. It's probably true in some cases, but people should ask themselves how the cost/benefit to calling it out looks.
For example, in one study, they divide participants into two groups, have one group watch https://www.youtube.com/watch?v=fn3KWM1kuAw (that highlights the high socio-emotional capabilities of a robot), while the other watches https://www.youtube.com/watch?v=tF4DML7FIWk (that highlights the low socio-emotional capabilities of a robot)
They are then asked if they agree or disagree with a (presumably hypothetical?) company's proposal to reduce employees' welfare, such as replacing a meal with a shake. Two groups showed a different preference.
This makes me think about that old question of whether you thank LLM or not. That is treating LLMs more like humans, so if what this paper found holds, maybe that'd nudge our brain subtly toward dehumanizing other real humans!? That's so counter intuitive...
Interestingly, both technologies also supercharge scams - one by providing a way to cash out with minimal risk, the other by making convincing human interaction easier to fake.