Without ethical AI, your company could be opening itself up to unnecessary risk.
Elon Musk’s firing of roughly half of Twitter’s workforce could lead to significant negative consequences for the company. As part of the mass layoffs, full teams were either eliminated or reduced to a skeleton crew. One such team was Twitter’s ML Ethics, Transparency, and Accountability team (referred to as the META team, actually), a group of around 20 people focused on making Twitter’s algorithms more transparent and less biased.
Twitter could not be immediately reached for comment. According to Axios, all but one member of Twitter’s communications team have been fired.
According to Ramayya Krishnan, dean of Carnegie Mellon University’s Heinz College of Information Systems and Public Policy, Musk made a grave mistake by firing the META team, which he referred to as a leader among the tech community in ethical AI, as machine learning plays a core role in keeping Twitter functioning.
But what exactly is ethical AI? Essentially, it’s the process of purifying algorithms to remove any kind of bias or unfairness. While Twitter is a company with an evident need for ethical, accountable machine learning, Krishnan says that any type of data-driven company could experience inadvertent consequences if they don’t thoughtfully implement machine learning. “If you were hiring for a software engineering role and used a faulty algorithm to scan all the resumes, it might look at past hiring trends and automatically disqualify the female applicants,” says Krishnan.
The META team’s main objective was to engender user trust in Twitter’s algorithms, and the team was given the latitude to publicly share their findings with the rest of the world. One of the team’s biggest accomplishments, according to Krishnan, was fixing the Twitter image cropping algorithm, which was found to be consistently cropping out black faces in photos posted to the app. Along with publishing their findings regarding the image cropping algorithm, The team published a study that found that right-wing-leaning news sources tend to benefit more from Twitter’s algorithm than left-leaning sources.
As for how the META team’s firing will affect Twitter, Krishnan says it’ll be less obvious than one specific problem. He added that not having ethical AI voices at the strategy table could result in Twitter having a more reactive posture when unintended issues pop up as a result of the company’s algorithms being half-baked at launch. With an ethical AI team, Krishnan says, companies can be proactive and thoughtfully test their algorithms before they are released and deployed.
Register for a free account and get Inc. Australia Today, our daily newsletter digest curated for entrepreneurs.