3 ms·

> Documents like these shouldn't be automatically processed, they should be reviewed by humans. Human review is fraught with bias. A lot of attention and colum

by tqi 5y ago

> Documents like these shouldn't be automatically processed, they should be reviewed by humans. Human review is fraught with bias. A lot of attention and column inches are devoted to algorithmic bias these days, but let's not confuse that for evidence that human review is perfect or even better. It's just harder to audit.

dahart 5y ago
I don't think anyone is confusing algorithmic bias for evidence that humans are better. It's the other way around -- the argument has always been that algorithms are biased because humans are biased. Having a systemic bias that masquerades as neutral seems worse than having people in the loop. When submitting my resume and interviewing for jobs, I'd certainly prefer to be subject to: 1) multiple people with different biases, 2) humans that can change their mind and/or recognize their biases, 3) humans that can explain their rationale, and 4) humans that can grasp and weight eccentric and non-conforming experience. I'm not interested in being ranked in a standardized way, because I'm not interested in being a pure cog in wheel. I also think people who want standardized resumes and automated processing are vastly underestimating the SEO effect it will have on hiring. If resumes are automated, then many people more aggressive than you will game the system to the Nth degree. The same thing that happens to Google and the internet will happen to jobs: the loudest and spammiest will win, and the good content will go mostly unnoticed and unrewarded.
plafl 5y ago
I think resistance to algorithms is about failure modes. If an algorithm becomes standard and you are an edge case then you are in trouble, much the same way you are in trouble today with companies that have automated customer support and make it hard to contact a human being. Losing your email account is annoying but not being able to have a job is worse. Automated systems are designed to avoid mistakes but are bad at correcting mistakes. I would argue that even with humans involved the hiring process is harder to some people, even discounting bias, just because it is more or less standard but at least they have hope in the variance that remains.

by tqi 5y ago

dahart 5y ago

I don't think anyone is confusing algorithmic bias for evidence that humans are better. It's the other way around -- the argument has always been that algorithms are biased because humans are biased. Having a systemic bias that masquerades as neutral seems worse than having people in the loop. When submitting my resume and interviewing for jobs, I'd certainly prefer to be subject to: 1) multiple people with different biases, 2) humans that can change their mind and/or recognize their biases, 3) humans that can explain their rationale, and 4) humans that can grasp and weight eccentric and non-conforming experience. I'm not interested in being ranked in a standardized way, because I'm not interested in being a pure cog in wheel. I also think people who want standardized resumes and automated processing are vastly underestimating the SEO effect it will have on hiring. If resumes are automated, then many people more aggressive than you will game the system to the Nth degree. The same thing that happens to Google and the internet will happen to jobs: the loudest and spammiest will win, and the good content will go mostly unnoticed and unrewarded.

plafl 5y ago

I think resistance to algorithms is about failure modes. If an algorithm becomes standard and you are an edge case then you are in trouble, much the same way you are in trouble today with companies that have automated customer support and make it hard to contact a human being. Losing your email account is annoying but not being able to have a job is worse. Automated systems are designed to avoid mistakes but are bad at correcting mistakes. I would argue that even with humans involved the hiring process is harder to some people, even discounting bias, just because it is more or less standard but at least they have hope in the variance that remains.