The tech industry promised to diversify last year. Has it delivered?

This week marks one year since a Minneapolis police officer murdered George Floyd. The year has seen national, even global, protests demanding justice for Floyd, Breonna Taylor and other Black victims of police killings as well as a racial reckoning in society and business. Many tech companies came out in support of the Black Lives Matter movement and made promises to make their companies more diverse and equitable in a notoriously nondiverse industry.

One year later, is it possible to measure their progress? I spoke with Megan Rose Dickey, a senior reporter at Protocol, where she covers labor and diversity in tech. The following is an edited transcript of our conversation.

Megan Rose Dickey: It’s hard to know exactly how well they’ve done because a lot of these commitments were looking at 2023 or 2025. So we’re still not going to see if companies have really followed through with those commitments for a little while.

Molly Wood: Yes, it is always somewhat convenient that it’s just a little farther out than we can measure.

Rose Dickey: Right. And they probably hope that we’ll forget about it.

Wood: Yeah, exactly. But there is culture. And we’ve seen a lot of stories about culture and attempts to improve that, which is harder to measure. But have you seen any actions that seem promising?

Rose Dickey: Yeah, there have been some actions. So I would point to Snap, for example. A big change that they made was around the use of racist language in their code database, so whether it’s like “master” and “slave” or “blacklist” and “whitelist.” And so they’re working on actually changing those terms. That’s one small thing, but I think that’s definitely a step in the right direction. And Snap has also had issues around the camera app engaging in digital blackface and things like that. There was this Bob Marley filter a couple of years back that basically was just blackface. And so they said that they’re working on that and working on kind of changing their algorithms and ensuring that it has more inclusive inputs. But then, I don’t know. You see Google. Last week, we found out that Google’s dermatology app barely used any data from people with darker skin tones. After all this time, how is that still what’s happening? It’s just kind of becoming comical at this point.

Wood: I wonder if some of these requests for inclusivity and improved culture and different, more inclusive code — are they being heard? Do you think that there are some companies who are hearing it and others who are just saying, “Leave us alone. We know what we’re doing”?

Rose Dickey: I think that there are some companies that are hearing it. But I think sometimes the issue can be that there can be a disconnect between the diversity departments and the broader organization. So while you might have a very ambitious, hardworking team of people that are focused on diversity and inclusion, that doesn’t always necessarily make its way over to, say, the engineering team or the executive team. And so I think that’s just why it’s so important that there’s this buy-in from the entire company, if they’re really wanting to see real change. But then, of course, you have companies like Coinbase and Basecamp and New Relic that are saying, “We don’t want to hear you talk about social issues, political issues, race issues. You just need to be very mission-focused. Focused on what the company is doing. And don’t expect us to really speak out about these types of racial or social or political issues.” And, I mean, I think that’s not necessarily to say that they don’t care about diversity. But I think there are some companies that are just like, “This has just become too much, and we just need to focus on the business itself.”

Wood: Are there any tech companies you can think of that have really handled the last year well in a measurable way?

Rose Dickey: I’ve been asked that question a lot just over the years of covering [diversity and inclusion], and I used to say, “Pinterest is doing really well.” And then, next thing we know, it’s actually not doing well. And so I think there are still definitely things that some companies are doing that look good on the surface, whether it’s like unconscious-bias training or ally training. But I think the bigger, more important things need to come around, like really overhauling the hiring practices, like where they’re recruiting from, and looking at the performance-review process and seeing if there are biases and inequities in those. Because a lot of times, [and] I’m thinking about Amazon right now, there have been some reports about people getting deleveled at the company, meaning that they’ll apply for a job at one level, but then they’ll be brought on at like a lower level or two levels lower. And oftentimes, these are people of color or they’re people from other underrepresented groups in tech. And so I think if we’re really wanting to see real change, then companies need to look at some of those more systemic processes that they have in place to really ensure that people, people of color and other underrepresented groups, are really able to thrive with these companies and feel like they have a voice and feel supported there.

Related links: More insight from Molly Wood

The most recent tech company to make a big public commitment to diversity is HP, which last week announced that it wants to become the “most sustainable and just” tech company in the world. It said it wants gender parity in HP leadership, to have more than 30% of its technical and engineering staff be women and to do as well or better than the tech labor market overall in terms of racial and ethnic diversity by 2030. We’ll keep checking.

Back in April, a small group of tech companies actually teamed up to advance diversity and inclusion efforts. The Alliance for Global Inclusion was formed by Dell, Intel, Nasdaq, Snap and NTT Data, focusing on diversity in leadership and executive boards, inclusive language and industrywide standards — the things Megan mentioned — and more inclusive product development around things like artificial intelligence.

Like remembering, in the year of our Lord 2021, that when you’re training your new dermatology app to use deep learning to identify potential skin conditions, that you should try to find a dataset that includes all kinds of different skin colors instead of one that uses more than 64,000 images of skin, only 3.5% percent of which are of brown, dark brown or black skin, according to a piece in Motherboard about Google’s new app.

Google says its models perform well across ethnicities and showed data to Motherboard showing that its system was more than 87% accurate at identifying skin conditions for Black patients. However, those results are based on ethnicity and not skin color, and no one with what’s known as Fitzpatrick skin type VI, dark brown or black, was represented in that data. 

In a statement to us, Google said: “Our peer-reviewed research has shown that our AI-model performs well across all ethnicities, and improves the ability of primary care physicians and nurse practitioners to interpret dermatologic conditions by more than 20%, including for Black and Hispanic / Latino patients. Equity has – and will continue to be – a key focus in the development of our technology, including working with datasets that are inclusive of different ethnicities, skin types and age groups, and partnering with clinicians and experts who have experience working with communities of color. We are committed to continued research and further developments as we help connect people to relevant information about common skin questions.”