Twitter to offer cash bounty to users who spot algorithmic bias
The bounty has been set at up to $3,500
August 01, 2021 / 01:08 PM IST
Twitter wants users to spot bias in its machine learning systems
Twitter has announced a bounty of up to $3,500 (approx. Rs. 2,60,000) for users who are able to spot algorithmic bias on the platform. The company has said that this bias bounty is the first of its kind.
Machine Learning bias or Algorithm bias happens when AI produces results that are unintentionally prejudiced due to an error in the assumptions it makes.
If a result has a high degree of bias associated with it, then the result is considered inaccurate. It can also be described as the machine's inability to predict the correct outcomes by not taking into account all the information that has been shared with it.
As you might have guessed, on social media platforms, these can have dire ethical consequences. Not too long ago, Twitter ran a machine learning based automatic image cropping program that turned out to have an unintentional racial bias.
Users found that in most cases, the AI would crop out the picture of black persons in preview. Later the company admitted that the tool showed an "unequal treatment based on demographic differences."
In a blog while announcing the new bounty, Twitter said that, "Finding bias in machine learning (ML) models is difficult, and sometimes, companies find out about unintended ethical harms once they’ve already reached the public. We want to change that."
The company said they have seen how, "research and hacker communities helped the security field establish best practices for identifying and mitigating vulnerabilities
" and the company wants to create a similar ecosystem for ML ethics.