In the summer of 2019, I had the privilege of launching the Purple Certification program in collaboration with Bo Young Lee, who at the time was serving as Uber’s first Chief Diversity & Inclusion Officer. Over breakfast with tech and policy journalists in New York City, Bo and I talked about our vision for a certification program that would recognize employers taking recommended steps to address sexual harassment and promote safety in the workplace.
Bo and I launching the Purple certification program in July 2019
Six and a half years later, I’m happy to report that the certification program is still going strong. As we prepare to launch version three (“v3”) of our employer survey in January, I’ve been talking frequently with workplace leaders about current workplace safety trends. As in many areas of business right now, Artificial Intelligence (“AI”) use is top of mind for employers.
That made it the perfect time for me to catch up with Bo, who now serves as the CEO of AI4ALL, a non-profit organization whose mission is to ensure that the next generation of AI leaders reflects humanity. Bo and I had a wide-ranging conversation about the intersections of AI use and workplace misconduct, and about how leaders can promote equity-centered AI adoption. I hope you find our conversation as enlightening as I did—and that it helps you think about how to promote responsible AI use in your own workplace, community, or organization.
***
Ally: This is a moment in which some organizations seem to be reevaluating their existing equity-related policies and programs. Why is it still important for employers to maintain strong anti-harassment policies and programs, and what should they look like?
Bo: Even before the AI era, I always told leaders that all workplaces need strong anti-harassment and discrimination policies, because humans are fundamentally fallible. Regardless of what kind of company you run, or how careful your hiring practices are, bad things will happen in the workplace. As a leader, your job is to: (1) try to prevent those instances from occurring in the first place, and, perhaps more importantly, (2) ensure that the organization handles incidents well when they inevitably occur, despite your best prevention efforts.
That means ensuring people can report problems safely and confidentially, without fear of retaliation, and that transparent, accountable processes are in place to handle those reports. Importantly, those processes should include some form of restorative follow-up, with communication both with the person who reported the incident and the person alleged to have caused harm, regardless of the outcome.
For example, in a sexual harassment case, even if the person who engaged in harassment is ultimately terminated, the employer should still reach out to the person who was harmed and ensure they are made whole. This might require additional steps to help the employee feel safe and professionally fulfilled–such as reconsidering project or team assignments–in the aftermath of the incident.
Ally: How has the introduction of artificial intelligence tools changed things?
Bo: To my earlier point about human fallibility, it’s simply a fact that people will inevitably use AI in inappropriate ways–both inside and outside the workplace. For example, 90-95% of all AI deepfakes are pornographic or involve sexual depictions of women and girls. This is already a serious problem in schools, where deepfakes are used for peer bullying or harassment, and it could also happen in the workplace, if it hasn’t already.
Imagine a scenario in which one colleague becomes obsessed with another colleague and creates a deepfake of them, or a companion chatbot designed to look or act like them. That could become a form of stalking or workplace harassment and raises difficult questions about employer responsibility. A deepfake created solely for personal use, rather than disseminated externally, may not be as clearly addressed under existing law–yet it can still reinforce harmful behavior and create serious workplace culture issues.
Ally: What are some steps employers should take to address these risks?
Bo: The more people engage with AI, the more they can detach from reality–and yet many companies are pushing employees to use AI tools without addressing the associated mental health risks. Organizations need to put audit and governance processes in place and create roles dedicated to ensuring responsible AI use.
If a company doesn’t have a dedicated role for this, I recommend making responsible AI use part of the Chief Ethics Officer’s portfolio. At Uber, I helped start both a Product Equity Team and an Algorithmic Fairness Team, which were focused on ensuring that our algorithms–themselves a form of AI–were not perpetuating bias or reinforcing discriminatory outcomes, such as race- or gender-based pay disparities. That’s a best practice any company using algorithms should consider adopting.
Ally: Those are great suggestions. Let’s say a company takes your advice and creates a new role dedicated to responsible AI use in 2026. What should that person be thinking about when deciding which AI tools to bring into the workplace?
Bo: First, companies should ask AI vendors to share their audits, bias-testing documentation, and fairness metrics to demonstrate the steps they’ve taken to address equity concerns. I also recommend asking for transparency around how a product’s algorithm works. While some elements may be proprietary, vendors should be able to explain the basics of their systems and share information about the sources of their data.
It’s also critical to ensure that vendors conduct ongoing testing for bias and fairness issues. Providing that information only at the point of sale isn’t enough to ensure the tool continues to evolve in ways that align with your organization’s values.
Finally, organizations should establish their own internal review committee–ideally including leaders from Human Resources (“HR”), Legal, and Information Technology (“IT”) –to develop metrics and processes for continually assessing AI bias in the workplace.
Ally: Thank you so much for sharing your wisdom with us, Bo!
***
Bo certainly gave me a lot to think about, and I hope this conversation sparked some questions or ideas for you as well. Is there anything you wish I’d asked Bo but didn’t? Drop me a note in the comments or the chat–I’d love to hear from you, and maybe even track down an answer.