3 ms·

No, it's that the average unpaid human doesn't care to read closely enough to provide signal to improve AI models. Not that they couldn't if they put in even th

by gpm 10mo ago

No, it's that the average unpaid human doesn't care to read closely enough to provide signal to improve AI models. Not that they couldn't if they put in even the slightest amount of effort.

ehnto 10mo ago
Why would an unpaid human want to do that?
- alterom 10mo ago
  Exactly — they wouldn't.
- 0manrho 10mo ago
  Therein lies the problem.
kazinator 10mo ago
Firstly, paying is not at all the correct incentive for the desired outcome. When the incentive is payment, people will optimize for maximum payout not for the quality goals of the system. Secondly, it doesn't fix stupidity. A participant who earnestly takes the quality goals of the system to heart instead of focusing on maximizing their take (thus, obviously stupid) will still make bad classifications due to that reason.
- tbrownaw 10mo ago
  > Firstly, paying is not at all the correct incentive for the desired outcome. When the incentive is payment, people will optimize for maximum payout not for the quality goals of the system. 1. I would expect any paid arrangement to include a quality-control mechanism. With the possible exception of if it was designed from scratch by complete ignoramuses. 2. Do you have a proposal for a better incentive?
  - Eisenstein 10mo ago
    1. Goodhart's law suggests that you will end up with quality control mechanisms which work at ensuring that the measure is being measured, but not that it is measuring anything useful 2. Criticism of a method does not require that there is a viable alternative. Perhaps the better idea is just to not incentivize people to do tasks they are not qualified for
- dresrs 10mo ago
  > Secondly, it doesn't fix stupidity. Agreed, and would add that it doesn’t fix other things like lack of skill, focus, time, etc. An example is the output of the Amazon Turk “Sheep Market” experiment: https://docubase.mit.edu/project/the-sheep-market/ https://docubase.mit.edu/project/the-sheep-market/ Some of those sheep were really ba-aaa-ad.
- zem 10mo ago
  I don't think there is any correct incentive for "do unpaid labour for someone's proprietary model but please be diligent about it" edit: ugh. it's even worse, lmarena itself is a proprietary system, so the users presumably don't even get the benefit of an open dataset out of all this

by gpm 10mo ago

No, it's that the average unpaid human doesn't care to read closely enough to provide signal to improve AI models. Not that they couldn't if they put in even the slightest amount of effort.

ehnto 10mo ago

Why would an unpaid human want to do that?

alterom 10mo ago
Exactly — they wouldn't.
0manrho 10mo ago
Therein lies the problem.

kazinator 10mo ago

Firstly, paying is not at all the correct incentive for the desired outcome. When the incentive is payment, people will optimize for maximum payout not for the quality goals of the system. Secondly, it doesn't fix stupidity. A participant who earnestly takes the quality goals of the system to heart instead of focusing on maximizing their take (thus, obviously stupid) will still make bad classifications due to that reason.

tbrownaw 10mo ago
> Firstly, paying is not at all the correct incentive for the desired outcome. When the incentive is payment, people will optimize for maximum payout not for the quality goals of the system. 1. I would expect any paid arrangement to include a quality-control mechanism. With the possible exception of if it was designed from scratch by complete ignoramuses. 2. Do you have a proposal for a better incentive?
- Eisenstein 10mo ago
  1. Goodhart's law suggests that you will end up with quality control mechanisms which work at ensuring that the measure is being measured, but not that it is measuring anything useful 2. Criticism of a method does not require that there is a viable alternative. Perhaps the better idea is just to not incentivize people to do tasks they are not qualified for
dresrs 10mo ago
> Secondly, it doesn't fix stupidity. Agreed, and would add that it doesn’t fix other things like lack of skill, focus, time, etc. An example is the output of the Amazon Turk “Sheep Market” experiment: https://docubase.mit.edu/project/the-sheep-market/ https://docubase.mit.edu/project/the-sheep-market/ Some of those sheep were really ba-aaa-ad.
zem 10mo ago
I don't think there is any correct incentive for "do unpaid labour for someone's proprietary model but please be diligent about it" edit: ugh. it's even worse, lmarena itself is a proprietary system, so the users presumably don't even get the benefit of an open dataset out of all this