Episode 28: AI Ethics, Bias & Business: Moojan Asghari, Co-Founder, Women in AI & Global Tech Entrepreneur

As global businesses integrate AI into processes, jobs, and public outreach, how can leaders ensure those efforts are conducted ethically and free of bias?

Transcript

Chris Riback: I'm Chris Riback. This is Call In with Dr. Alexandria White. We discuss business leadership in our time of social change when to call in, when to call out, and how to build sustainable business value today.

Before our conversation though, an ask from us to you. We hope you like these call in conversations. And if so, we'd appreciate if you take a moment, go to Apple Podcasts or wherever you listen, and if you're so moved, leave a five-star review. The ratings really matter. They go a long way to helping other people find the podcast.

Dr. Alexandria White: Our show is brought to you by Clayton, Dubilier & Rice, which is committed to a more diverse and inclusive future. Let's call in.

Chris Riback: Moojan, thanks for joining us. We're excited to get to talk with you.

Moojan Asghari: Thank you so much. Yes, I'm excited too.

Chris Riback: Alex and I have been looking forward to it. So before we get into the topics at hand, AI and ethics, AI and bias and the things that leaders need to know and think about and act about in these areas, tell us about you, because you've got a wide-ranging background and those inputs obviously must help drive your thinking and your work in these areas. So tell us a little bit about your background, if you would.

Moojan Asghari: Sure. So I started my career in investment banking. My academic background is industrial engineering and management and corporate finance banking. Then I started working in a bank in Paris, but very soon, I realized what is passionate to me is more contributing to community causes. And that's where I started to work at different startups, tech startups mostly. And eventually, I went down the rabbit hole, we call it, and getting to know AI and working with some startups. And eventually, I created my own nonprofit organization with two of my co-founders, called Women in AI. And so we realized that there's this huge gender gap in the field and we started to understand what that implies in terms of the problems that it creates.

Dr. Alexandria White: Okay. So there's so many things that I want to talk about based off of that answer, but first inclusion and biases. So please talk about biases in AI.

Moojan Asghari: When we're talking about bias in AI, we're talking about a bias in decision making. That's literally what we really talk about when we're talking, when we say bias in AI. And the problem that we are facing is that AI acts as a little child that absorbs the basically knowledge, the biases of its creators. So most of the people who are creating these technologies today in the world, and the persona we can create for them is basically white engineers who are mostly male. And they are creating these technologies based on what they know, based on their culture, based on what is their ethical framework. And then that results in a bias result that doesn't respond to the global community of users. Let's say a woman in an African country doesn't necessarily benefit from the same thing of somebody who's a white man in let's say United States.

Dr. Alexandria White: Tell us a little bit about how AI can be more inclusive.

Moojan Asghari: So one of the ways that we can make sure that AI is not bias is, we need to make sure that we actually know that there's bias. So we are aware of that and we have the will to change that. So there's a very important shift in the mindset and to make sure that we understand and we acknowledge that. And then we want to take actions. So that means that we need to educate young children at school already, especially women, especially BIPOC communities, to help them to know that this is a path for them as well and learn about very, very technical aspects of AI. That's important because in order to develop AI, you need to have an engineering background. And so that is important to let them to know that they can be also part of this building communities.

Another way that we can do that is educating also our leadership teams to make sure that not only we are hiring the right talents to come in to increase the diversity of our teams, but also to maintain them. And that was actually a case of myself in a few of corporations I worked at, we hire women, we hire Black people, people of color, indigenous people, but the problem is that we lose them also because the environment is not ready to receive them, to maintain them, to make them feel safe in their workplace. And that actually doesn't work and it costs a lot to companies. So that leadership mindset shift, it needs to happen to create the safe environment. And the last thing I would say is to know how we are measuring the success for our companies.

If we want to receive to an inclusion inclusive AI, we need to make sure that what kind of KPIs we are measuring. Do we know actually if we are asking the right question to know if the AI is biased or not? Sometimes it's very even difficult to understand if the application is biased. Maybe the tests that they needed to be happening, they have not been happening and we just released the product and we receiving a lot of problems. That was the case, for example, for facial recognition applications and many other AI applications that has happened. So also measuring that, that is very important.

Dr. Alexandria White: Wonderful.

Chris Riback: So Moojan, talk to us about some of the challenges, what can go wrong, and for a company. I can understand, okay, we don't want unethical code being written. We don't want our automated functionality. And obviously AI and artificial intelligence, we don't want that to introduce bias in terms of outcomes, but literally what would it mean to have a biased outcome? What would that look like?

Moojan Asghari: Yes. I can give you an example. The issue with AI is that you don't even know behind the screen of the page that you see today, let's say you're going to find a job you want, you're applying for jobs. What I see as opportunities for me, because my data is all around internet in all my cookies in every account, and they are used in ways that I can't even control it today. So the data privacy and identity security is getting very, very out of control for the user because we are like the product for these applications that giant tech companies are using and paying millions and millions of it to get them. I don't even know there is an AI algorithm on the background to give me something that I might not necessarily be interested in or I'm losing an opportunity.

One example is, and that was actually the Google research in June, when you search for, let's say a job, if you're a woman, you would not mainly see the job offers more than 200K salary per year, but if you were a white man or a male in general, mostly white men, but men, you would receive those higher salaries because it was running on this stereotype of the society that women don't have to work too much and women don't have to, let's say gain a lot. So that's the assumptions that are coded by engineers on the background, but we don't know that. And the engineers, they are the ones that are feeding these data as an assumption and, of course, their biases to the platform. So that's, for example, one example can impact us.

Dr. Alexandria White: And so you talked about education, you talked about educating the young, so mistakes like this wouldn't happen. You said something about salaries. How can AI help with this gender gap in salaries and helping decrease the feminization of poverty? How can AI help with that?

Moojan Asghari: Yes, so there are actually some companies working on flagging the biases in AI programs themselves and then notifying and then trying to fix them.

Dr. Alexandria White: Okay.

Moojan Asghari: I don't remember the names, but there are a few startups. So they actually use AI to flag the AI biases and fix them. It also depends on what kind of programs we're talking about, but that's one way that AI itself could contribute to solving AI biases. But the main thing is not even coming from AI, is from humans to understand that thus, there's something to do. It's not easy because we are today facing a huge skill gap in the world. It's not even about women to be in AI, it's about engineers who are capable of working in the field, the data scientist, regardless of their gender, because the technology is advancing so fast. So the cycle of advancement of AI is very fast, that we are, just this gap of skills is getting bigger and bigger every year. So that's a global, actually crisis we have.

Chris Riback: Moojan, what's the connection between whom you hire to work on the development side, whom you hire to work in your technology teams, and the output of the code? Does diversity and engineering matter, or is it just all ones and Os and bits and bytes and the diversity of the engineers is less important than their actual capability and skills?

Moojan Asghari: Actually, there is more and more conversations about who you should hire in your development team. And you have development team, you have auditing team and so on. And the profile of a perfect developer is changing a lot. For example, we talked today about developers who know a little bit of ethics and maybe combined with even philosophy, or you're really shifting this because the more we go forward, ethics in AI is not a technical thing. It is a very philosophical thing. If you check out what would mean ethical 3000 years ago, that was completely different than what we see today. And so fairness or ethics, it changes over time with us humans that we grow in our knowledge and consciousness. So it is not a fixed thing. And many companies are trying now to combine different profiles in diversity. It goes beyond gender, it goes beyond race. It is really about combination of skills and combination of soft skills and hard skills together to make it a perfect, let's say, team for developing AI.

Dr. Alexandria White: Can you give us some success stories? Any examples of organizations, nonprofits that have made a significant progress in promoting diversity with using AI as maybe a foundation?

Moojan Asghari: That's a good question. Yes.

Dr. Alexandria White: And we know you deal with a lot of people, building companies, but any good example.

Moojan Asghari: So at Women AI, we had the opportunity to work with an innovation agency in Sweden. We conducted basically a report on how could AI contribute to various problems in the society, such as gender based violence in inequality of pay, that work was a foundation of something that they're taking it and they're working with different clients and partners that they have to advise. Them how to move forward with their strategies. I think that there is not a perfect answer to this. I think what we need is collaborating together. I know for example, I was involved with some of the European Commission hearings, public hearings and meetings that we have to come together and work on the ethics of AI and different aspects that come in the conversations that we have to tackle certain issues with AI or about AI.

And all of think, I think it's an ongoing work. I think that, that mean we have example that can go very wrong, but what is working today is those companies that they have flexibility of shifting and changing and adapting to the advancements of the technology. And they have a good company culture that is a culture of openness and being able to accommodating diversity in general.

Chris Riback: Moojan, do you have a sense of how this is being addressed, whether it's through regulatory measures, whether it's through business councils? Regionally around the globe, you're located in Europe, you have contact and interact with other AI and women through your Women in AI organization with companies and engineers and women leaders around the world. Do you have a sense of how this is being addressed in the US versus EMEA versus Asia versus LATAM?

Moojan Asghari: Yes, so I think actually Europe has a very strategic position. For example, there were the initiatives of GDPR that slowly got adopted in different forms, in different parts of the world. They also, again, initiated the frameworks of ethics of AI. And so how it works normally with the commission is that they come together, they do their research and report and all these meetings and hearings and they become first recommendations. Then it becomes guidelines and then it becomes regulations. So that happened with GDPR and data privacy. Now it's happening with AI and they actually are adopting all of them, and eventually they force other geographies to follow that because otherwise they can't operate in the EU region. So I think it's very interesting this, it protects the end users and it avoids problems that might happen because of some companies or their mistake without knowing.

So that's something that is, it's happening. And I think Europe is a pioneer in that. And I know for example, slowly every single country, they started to have their own AI regulations and understanding of what does it mean ethics in AI. But more and more you see that they share the same things and companies also to do that. For example, one of the pioneering companies that they came with the AI principles was Accenture. And then slowly, many other companies, they have that. And it's becoming something as a standard, like having a code of conduct.

Dr. Alexandria White: For our listeners, as we said in the introduction, we have business leaders, C-suite, just people who are trying to do better, be better inclusive leaders who are our listeners. What are three things leaders should consider in regards to AI?

Moojan Asghari: I think the first question they need to ask themselves is why do they want to use AI? So that's a very important question because you see many companies, I think we are AI powered, we are AI first, et cetera. And that sometimes can just be a marketing angle and sometimes they actually can make them a bit go farther from their main mission and their main value proposition. So that's a very important question that why do we want to use ai? Is it a need or is it a want or what is really that? And then it translates into the strategy of breaking down the different phases of it to adopt it.

Another important thing is that once they have the vision and they know, for example, what they want to do with AI, which parts they want to develop is the how? And how is very important too because it includes humans, it includes processes, the humans, they need to be, let's say, diverse, these profiles, these geographies and so on. The processes means that how are we making sure that we flag issues, we iterate the programs, we make sure that it's in a safe container and they're rules and responsibilities, who's accountable if something happen and so on. So these two, I think are the most important thing in order to reach to inclusive AI and actually serving the people who are using our products and services in the most efficient way.

Dr. Alexandria White: Thank you. Thank you for that.

Chris Riback: Moojan, I think you get to the heart of the question around what is and why is there bias in AI? And what you were just discussing was investigated a bit by the McKinsey Global Institute back in 2019, and the line which relates I think to what you're saying, and I would love to hear your point of view on it. They write, many experts tend to welcome algorithms as a refreshing antidote to human biases that have always existed, at the same time, many worry that algorithms may bake in and scale human and societal biases.

So on the one hand, we all look at technology and we're like, yay, technology. It's going to disaggregate human intention, whether that's for good or for bad. It's going to disaggregate that human intention out and almost anesthetize the outcomes because it's technology, it's not human, it's not sentient. And yet on the other hand, it's created by humans. And we all bring our own backgrounds. And I know you said, I know you have talked about the philosophy around what it is that we're discussing in this challenge, but how do you think about that apparent contradiction between we're talking about technology which necessarily removes emotion, removes and humans, who are naturally emotional entities?

Moojan Asghari: You're tackling actually a topic that is my passion, is how could AI serve humanity and how could it make us basically more human? Which is, I think that's the purpose of AI. We can definitely talk about that. But talking about bias again in AI, we are, all of us, bias as individually. That's a reality. You are biased. I am biased. Alex is biased. All of us, we are biased because we are born in a special environment. We were educated in a different way. We speak a  different language, we have different understanding of the world. That is why we are each of us biased. And when we say biased is basically the, let's say we have this middle line in the world, and that's the place that we want to say everybody can benefit from what we are building, but then everybody is different.

So how we could reach to that middle ground that everybody would be happy and would be served and would be treated equally. And that's very difficult. But to know that we are all biased by collectively bringing our consciousness, our intelligence together, we could make that happen. And that's what I love actually when I go to the blockchain world and this decentralized autonomous organizations mindset that people are now more and more building. You see how eventually AI and our technologies could be more inclusive by including more people and giving them the power to be builders and decision makers.

Dr. Alexandria White: I love that answer, bringing people together, we're more like than we are different. The diversity of thought, I really, really agree with that sentiment. Is there anything else that we missed that you'd like to cover that? Is there something impactful that you'd like our listeners to know about Women in AI?

Moojan Asghari: I would say about Women in AI, we are a global community today in 150 countries, purely on a voluntary basis is free for everyone to join. We say Women in AI because initially started to serve women, but we have so many men as well because we need their support-

Dr. Alexandria White: Correct.

Moojan Asghari:...they are our cheerleaders and supporters. So everybody is welcome. Please join us, help us to make a more inclusive world with the help of AI developing that together.

Dr. Alexandria White: Thank you. We will have that information about Women in AI in our show notes for listeners.

Chris Riback: Moojan, thank you. Thank you for taking the time to help us think through and understand AI and bias, ethical AI.

Moojan Asghari: Thank you for your invitation.

Dr. Alexandria White: Thank you.