Twitter Opens A Competition To Solve Problems With Algorithms That Are Considered Discriminatory, How Much Is The Prize?

JAKARTA - Twitter is offering users cash prizes that can help eliminate bias in its automatic photo cropping algorithm. The social media platform announced a 'gift' of $3,500 as part of this week's DEF CON hacker convention in Las Vegas.

The San Francisco-based social network has faced criticism that its 'saliency' algorithm, which uses artificial intelligence to decide which parts of the larger image are most relevant or important to preview, is discriminatory. Because their technology is considered to objectify women and choose white subjects over people of color.

“Finding biases in machine learning models is difficult, and sometimes, companies find out about unintended ethical harm once they reach the public,” Rumman Chowdhury and Jutta Williams of Twitter's Machine-Learning, Ethics, Transparency and Accountability (META) project said in a statement. a blog post.

"We want to change that. We want to grow a similar community, which focuses on ethics (machine learning), to help us identify a wider range of issues than we can do on our own," the post added.

In a post last May, Chowdhury said internal Twitter testing found only a 4 percent preference for whites over blacks, and an 8 percent preference for women over men. "For every 100 images per group, about three are cropped at locations other than the head," he added.

But Chowdhury, who has a background in algorithmic auditing, said Twitter was still concerned about possible bias in automated algorithms because "people are not allowed to represent themselves the way they want on the platform."

“Saliency also harbors other potential dangers beyond the scope of this analysis, including insensitivity to cultural nuances,' he wrote.

According to the instructions on the HackerOne site, the submitter doesn't have to be in DEF CON to enter this week's contest.

Entries will be judged against a variety of weighted criteria, including whether there was any defamation, stereotype, underrepresentation or misrepresentation, deletion, or reputational, economic, or psychological harm to the subject - and whether the harm was intentional or unintentional.

In their competition guidelines it was also stated: “The allocation of points is also intended to incentivize participants to explore representational hazards, as they have historically received little attention”.

The contest runs until Friday, August 6, at 23:59. The challenge also promises a second place prize of $1,000 and third place $500, with $1,000 each for the most innovative submissions and applicable to most types of algorithms.

Winners will be announced at the DEF CON AI Village workshop hosted by Twitter on August 9.

As an industry 'first', Williams and Chowdhury said they hope the contest sets the standard "for the proactive and collective identification of algorithmic hazards."