Twitter Launches The Algorithmic Bias Bounty Challenge
Twitter has announced a new bounty competition with cash prizes of up to $3,500 for revealing biases in its automated image crops.
HackerOne will award cash prizes to the winning teams: $3,500 for first place, $1,000 for second place, and $500 for third place. Most Innovative and Most Generalisable each receive $1,000.
"In May, we shared our approach to identifying bias in our saliency algorithm (also known as our image cropping algorithm), and we made our code available for others to replicate our work," said Rumman Chowdhury, Twitter's Director of Software Engineering, in a blog post published on Friday.
"We want to take this work a step further by inviting and incentivizing the community to help identify potential harms of this algorithm beyond what we've already identified," Chowdhury said.
They are influenced, as per Chowdhury, by how the research and hacker groups assisted the security field develop best practices for defining and alleviating weaknesses to safeguard the citizens.
"We want to build a community focused on machine learning ethics to assist us discover a broader range of challenges than we could on our own," Chowdhury added.
"With this challenge, we hope to establish a standard for proactive and collective identification of algorithmic harms at Twitter and in the industry," Chowdhury noted.
Twitter stated that for this challenge, they will re-share their saliency model as well as the code used to produce a crop of an image given a predicted maximally salient point, and participants will be asked to create their own evaluation.
In their approach, winning entries will take into account both quantitative and qualitative methods.
To submit a valid entry, all participants must sign up for a HackerOne account; anyone with a HackerOne account is eligible to compete.
The victors will be declared at the DEF CON AI Village workshop organized by Twitter on August 8, where they will be asked to showcase their work.