Diversity in AI leads to better AI

Are you sitting? I’m about to give credit to corporate suppliers. Yes, there was a lot of hype around AI at the keynotes this spring, but behind the scenes, more often than not, I heard details about how to achieve more responsible (and effective) AI.

On the other hand, I haven’t heard enough about the problem of bias – bias in training data, bias in outcomes, and bias in the humans who develop AI. Admittedly, bias is an extremely difficult problem – and one that is never fully solved.

I also haven’t heard enough about diversity in AI teams. My research to date indicates that more diverse teams simply create better AI. And they tend to detect bias earlier and faster. We can define “diversity” in many ways, beyond legally protected (and generally underrepresented) groups. Here’s the interesting part: Diverse teams tend to ask different questions about AI—and push AI projects in new and fruitful directions. As Aaron Harris, CTO of Sage, told me:

Obviously, if you’re going to teach AI to make decisions like a human, you have to represent the whole of humanity in that process. What we don’t often think about is the decision around which we’re going to decide: What problems are you going to solve? When you have a more diverse set of people leading the development of AI, you’re going to have a different set of problems that you’re going to solve with AI. It’s broader. And I think that’s just as important as addressing bias issues.

Better AI requires diverse talent – ​​how Sage’s community partnerships fit in

At Sage’s analyst event in Atlanta last week, the unexpected happened: diversity in AI took center stage. Not in a bossy way, but in a community way, building on Sage’s larger investment in Atlanta itself (Sage is moving its North American headquarters to the Ponce City market area). Our discussion was sparked by Sage’s ongoing partnerships with the BOSS Network and Morehouse College. Harris told the assembled analysts:

If we’re really going to build a great organization in Atlanta, we have to have a multigenerational commitment to the community, where people know who Sage is, where students at these great universities see us as a place to work, where we’re visible in civic activities in the city… It’s two things: it’s the commitment to Atlanta for the long term. But it’s also a real recognition that as technology advances, particularly with AI, we need to have a more diverse talent pool.

These are laudable goals, but it’s all in the execution—and that’s where brands tend to stumble. Because execution is everything. That’s where you’re likely to find yourself under fire; that’s when we find out what a brand is really made of. Amy Cosgrove, Sage’s vice president of people for North America, kicked off the day by detailing notable partnerships in Atlanta and beyond, including two programs with Morehouse College. One of those programs is the Morehouse-Sage AI partnership. Amy Cosgrove:

Morehouse College is a historically Black college. It is part of the Atlanta University Center Consortium (AUCC), the world’s largest consortium of historically Black colleges… In 2023, we launched this first program, which is our AI program. We launched it so that we could work with Morehouse to address the low representation of Black employees in computing and technology.

This is a comprehensive program designed to help students understand the foundations of generative AI. It’s a 15-week course; over 20 AUCC students took it in the first year… Currently, four of the program participants are Sage interns this summer; they’re working for us in an AI and Cloudops co-pilot.

Harris then made a fiery comment about How Homogeneous AI Won’t Lead Us to Good AI. I thought that comment framed the whole day. As he told me later:

I’m always pleasantly surprised by how interested people are in this… They really care about our collaboration with Morehouse and the BOSS network. The other organization that we work with, and I’ve been working with longer than either of those two, is Teens in AI. Teens in AI was born out of a UN mandate. The sole purpose of Teens in AI is to encourage girls to pursue careers in AI. They prioritize countries where girls have the most challenges. It can be infrastructure, but it can also be culture—where it’s hardest for a girl to have a career in AI, and they host hackathons all over the world.

Want to think outside the box when it comes to AI? These AI hackathons for teens are a great place to start:

I often get to judge these hackathons, and it’s really impressive. Where their minds are, what they’re thinking, and what drives them to want to solve these hackathons.

The BOSS Network – Black Women Entrepreneurs Overcoming Obstacles

Speaking of The BOSS Network, they were there too. A few years ago, Madeline Bennett wrote about The BOSS Network’s Sage partnership for diginomica, with an update this week: Diversity Backlash Makes It Even Harder for Black Women Entrepreneurs to Get Funding). While in Atlanta, BOSS Network founder Dr. Cameka Smith was in attendance at a celebration for a BOSS Network member whose company was recently acquired by Proctor and Gamble for $800 million.

It’s a great story, but it’s not the whole story. I asked Smith about the statistics in Bennett’s article. Perhaps the most disconcerting? While Black women start more companies than anyone else in the United States, they received just 0.34% of venture capital funding in 2021. Smith responded:

The funding isn’t there, the support isn’t there, so we decided to invest in ourselves… If you look at the numbers, less than 1% of Black women get venture funding. I probably know half of those founders myself. I think for us, as a community, we’ve always understood that there’s systematic racism. So we took it upon ourselves to say, “Hey, we don’t want to focus on what’s not available, we’re going to take what we have and move forward.” That’s the spirit of an entrepreneur: you get very little – and you do something with it – and that’s the success story of a lot of entrepreneurs.

Can Sage’s larger investment in Atlanta help drive that success? That’ll be something to watch. As Harris told me:

In the cities where we have located our flagship offices, we wanted to be deeply connected and integrated into the community. Those two cities are Newcastle (UK) and Atlanta. In Newcastle, we have a much larger population than Atlanta… Newcastle is a much smaller city. So we deliberately chose to build our office in Ponce City Market because it is in the heart of historic Atlanta. And we chose to work with these organizations that directly reflect our purpose as a company.

Martin Phelps, director of global analyst relations at Sage, decried the lip service that is sometimes given to these issues. A shuffle in the speaker lineup prompted Phelps to say:

It actually worked really well because the points that I was going to make, Amy made them much more impactful than I could have. Those points were about our goal, which is to “break down barriers so everyone can thrive.” Generally, that’s about democratizing systems that allow small businesses to have the same kind of access that large companies have. But it also includes our goal in terms of our involvement in organizations like Morehouse and BOSS Network. There’s real stakes here. It’s not just about writing a check.

I’ve worked in other organizations where everyone had a corporate social responsibility program. But some of them are really just checks. I think there’s a difference here. We’re really embedded in those kinds of activities and organizations.

My opinion

The only way for Sage to prove its difference is to make it real. But I will say this: It’s unusual to start a business analyst day with community speakers. It’s rare to see diversity investments at the heart of AI discussions. Phelps is right: for this to work, companies really need to integrate. As Smith told us:

Every month, Sage brings in a mentor from the leadership team, and they’re honestly the shining stars of this program because they come in as mentors; they coach these women, and they feel so connected that this brand really believes in them, and goes the extra mile, to actively support them with real ideas and tactics.

Of course, diversity isn’t the only challenge facing AI. For those interested in diving deeper into Sage’s AI strategy, check out my Sage Transform article, “Building AI Software Is Completely Different” – How Sage Plans to Deliver Enterprise AI with Thousands of Customer-Specific Financial Models. During our interview in Atlanta, I asked Harris to expand on his statement that “Customers need custom AI models, which is a much bigger challenge than multi-tenant.’ He has answered:

Building AI solutions is fundamentally different than building traditional software. The best way to describe it is to say that you are building a factory that automates the process of training, deploying, operating… coding the infrastructure East the work. Then you have your data scientists; they figure out what are the best algorithms, what are the best approaches to organizing the data, but then basically what you have to do is build a factory that automates all of that and manages all of that, and that’s the bulk of the work.

Today, it’s impossible to connect AI infrastructure coding to AI team diversity, but here’s the short version: If you want to build better AI, data management and algorithm tuning are a good place to start. Democratizing these skills is a worthy mission. When I met Dan Miller, Sage’s Executive Vice President of Finance and ERP, I told him that at this event, Sage’s mantra of “Break down barriers” seemed to be the reality. According to Miller, “Break down barriers, so everyone can enjoy” has evolved and taken hold.

It’s something that has stayed with me. Almost every time I stand up in front of a group of people, I say it — not because I feel compelled to say what our statement of purpose is, but because it’s at the heart of who we are and what we do for the people we’re trying to serve.

You can also view this statement as a challenge or a high bar – and that’s a good thing.