I dont know what magical humans you interact with. But the claim someone else doesn’t have a bias you may not be aware of… is fucking wild. People are wildly complex and the chances of you even REMOTELY understanding a fraction of their biases or thought processes with out extreme amounts of exposure and discussion with them is comical at best.
An LLM is infinitely more predictable and obvious in its biases then a human.
I will 100% agree on the people are less likely to just agree with you. Unless they really do not give a shit about the work at hand but then its really obvious.
The biases I’m talking about here are things like their generally preferred approach to solving problems, or how optimistic or pessimistic their outlook on certain classes of problems are. These are things you tend to know about the people you’re likely to brainstorm with, plus you can also be fairly sure that you’re all working towards the same goal. When you’re using an LLM like a partner to do this you can’t know those biases, as they’re externally controlled, you can’t know how they’ve changed since the last time you interacted with it, because the training data may have been updated, and it’s definitely not working towards the same goal as you because it’s not working towards anything, it’s just following statistical patterns in its training data.
I dont know what magical humans you interact with. But the claim someone else doesn’t have a bias you may not be aware of… is fucking wild. People are wildly complex and the chances of you even REMOTELY understanding a fraction of their biases or thought processes with out extreme amounts of exposure and discussion with them is comical at best.
An LLM is infinitely more predictable and obvious in its biases then a human.
I will 100% agree on the people are less likely to just agree with you. Unless they really do not give a shit about the work at hand but then its really obvious.
The biases I’m talking about here are things like their generally preferred approach to solving problems, or how optimistic or pessimistic their outlook on certain classes of problems are. These are things you tend to know about the people you’re likely to brainstorm with, plus you can also be fairly sure that you’re all working towards the same goal. When you’re using an LLM like a partner to do this you can’t know those biases, as they’re externally controlled, you can’t know how they’ve changed since the last time you interacted with it, because the training data may have been updated, and it’s definitely not working towards the same goal as you because it’s not working towards anything, it’s just following statistical patterns in its training data.