Responsible and Sustainable AI

Transcription

David Stearns:
Welcome to BSR Insights. I'm your host, David Stearns. In this series, we'll be talking to BSR experts from across our focus area and industry teams to discuss the environmental and social implications and opportunities arising from the increasing development and deployment of artificial intelligence in the global economy.

We're joined today by Lale Tekişalp, Associate Director, technology sectors at BSR, where she works with BSR member companies to incorporate human rights into business operations and strategy. She specifically focuses on the intersection of human rights and technology and also leads the Tech Against Trafficking collaboration. So welcome, Lale.

Lale Tekişalp:
Hi, David. Thanks for having me.

David Stearns:
It's great to have you here. So to kick us off, can you get us caught up on some of the latest developments, including some of the recent regulatory developments, as well as any of the guiding frameworks developed by various international bodies that our listeners should be aware of?

Lale Tekişalp:
Yeah, definitely. As you know, AI is having quite a moment right now, even though the technology has been around for decades. This is partly due to the developments that we've seen around generative AI technologies in the last couple of years. The release of ChatGPT and other generative AI tools that are powered by large language models have made AI much more widely adopted by consumers around the world. And similarly, businesses have started adopting AI at a much faster rate and making significant investments in new AI tools. It's almost become a race among companies, not only in the tech sector but beyond to use more AI and position it as a competitive advantage. It's almost like if you don't talk about AI these days, you're behind or becoming obsolete as a business. So in such an environment where there is so much commercial incentive to rapidly deploy this technology, there is not enough consideration of the risks that it might pose.

And I don't think there is a lack of awareness that AI comes with risk. There's been lots of public debate about AI safety issues, about privacy and copyright concerns, about the existential threats of AI. We hear about these a lot in the media, but I think we can easily say that businesses are largely unprepared to address the risks of AI. On the other hand, there is more and more expectation, as you said, for companies to consider these risks and to build and deploy AI in a responsible way. Many governments, international organizations, industry bodies have put out guidelines and standards on the responsible use of AI over the last few years, and there are new regulations that will impact companies.

Perhaps the biggest one of these is the European Union's AI Act, which is a new piece of legislation that will regulate AI in the EU, but it will impact companies beyond just the EU. We have another episode of this podcast series where my colleague Richard will speak about AI regulations in more detail. So I would encourage our listeners to listen to that if they're interested.

David Stearns:
Thanks for that, Lale. That's a great way to kick us off. You mentioned as companies are competing to pick up this technology as a competitive advantage, and this of course has been accompanied by a growing recognition of some of the risks that are associated with it. What are some of the specific environmental and human rights impacts posed by generative AI that companies should be aware of when designing, developing and deploying these technologies?

Lale Tekişalp:
Yeah, so I'll start my answer with a caveat. The impacts of AI systems depend very much on the context in which they're built and deployed. So the risks of AI technology deployed at a school in rural India, for example, will be very different than the risks of AI deployed for criminal justice in the US, which will be very different than the risks of AI deployed by a bank to calculate credit scores or AI deployed by a hospital to diagnose patients. We have to understand the specific use case and the specific local context to be able to discuss the impacts of AI, and companies can do this best by engaging with relevant stakeholders and impacted communities. That being said, I think we can talk about some general headlines to establish a starting point for companies. When we talk about AI and sustainability, I think there are four categories of impact that we should consider. Impacts to people, impacts to the environment, supply chain impacts, and longer term societal impacts.

Starting with impacts to people, we know that the use of AI by businesses can affect people and their rights in many ways. We typically list bias and discrimination at the top of these human rights issues. Let me give a few concrete examples. So when companies use AI systems in hiring or firing decisions, for example, that may result in the discrimination of underrepresented groups. Or when banks use AI systems to determine credit scores, again, this may result in the discrimination of marginalized communities or people who are already excluded from the financial system. Because AI systems are typically trained on historical data, so they reinforce existing inequities that shape the context in which these systems are deployed.

But bias and discrimination are not the only human rights risks related to AI. There are significant issues related to privacy and surveillance. For example, we are seeing more and more examples of companies using AI to monitor their workers and to analyze their behavior in the workplace. This not only has impacts to workers' right to privacy, but it can also impact their right to just working conditions, their right to freedom of association and many others. In fact, AI can impact every single human right depending on how it's deployed, but I'll leave it there because I know my colleague Hannah talks about the connection between AI and human rights in much more detail on one of the other episodes.

When we turn to environmental impacts, this is a question that many corporate sustainability leaders are struggling with. How to think about the environmental impacts of AI. Is it going to increase our company's carbon footprint or will it lead to efficiencies that offset the increase in computing power? The truth is it's not black or white. And I would recommend our listeners check out the episode on the environmental impacts of AI where my colleague Amir discusses this in more detail.

When we're considering these social and environmental impacts, it's important to think about the full AI value chain. I mainly talked about the downstream impacts of AI that are related to its use, but there are also sustainability issues related to the supply chain of AI. For example, there are labor rights violations in the AI data supply chain that many companies are not aware of. Data enrichment workers or click workers are behind many of the AI models we're using, and we know that their rights are being abused.

And finally, there are also broader, longer-term societal and economic impacts of AI, such as implications on employment and labor markets or implications on democracy and information ecosystems. And I encourage our listeners to check out the episode by my colleague Jacob on how to apply creative approaches such as futures thinking to understand and deal with those as a company.

David Stearns:
So thank you for that, Lale. And you've clearly laid out that some of the issues that are raised by the deployment of AI, they come with a lot of complexity, certainly a lot of nuance in terms of, as you mentioned, some of the context. You have to understand the context in which these impacts are being felt. Thinking about now what business leaders can start to think about and do to address some of these, I know BSR recently put out a guide on the responsible and sustainable use of AI, and you and your colleagues laid out a number of steps that businesses can look at as a starting point for addressing this. And there were five specific actions that were laid out, starting with adopting a responsible approach to the development and deployment of AI, undertaking human rights due diligence, assessing environmental impacts, developing responsible sourcing approaches, and finally applying futures thinking to explore other impacts of AI. As you've mentioned, we're going to dive deeper into some of these throughout the series and conversations with some of your colleagues, but perhaps you can kick us off by providing us a broad overview of what these steps mean. What are they and why are they really important for business leaders to consider?

Lale Tekişalp:
The most important first step we recommend to companies is to understand the impacts of AI systems deployed across their business. This allows companies to create meaningful and targeted approaches to responsible AI. We're seeing many companies putting out responsible AI statements and policies these days and set up AI governance structures. These are great, but without fully understanding the impacts of AI, these policies and governance bodies will not be very effective.

So in order to understand the impacts of their AI systems, companies can do a number of things such as conducting human rights assessments or integrating AI into their materiality assessments. Once we have a clear understanding of the impacts, then to manage these, companies should make sure they're integrated into existing sustainability management approaches such as human rights due diligence processes or climate strategies or responsible sourcing programs. It's really helpful when companies establish responsible AI programs. These are typically cross-functional efforts including sustainability, human rights, legal compliance, but also data and AI teams, engineering teams, sales or marketing teams, depending on the company's AI use cases.

And finally, of course, it's helpful to have policies, principles, governance structures to make sure AI is built and used in a responsible manner. It's also important to note, I think that these actions don't only apply to tech companies or to the developers of AI. These sustainability issues are relevant for every industry, every company that deploys and uses AI technology. So every business needs to think about responsible AI and understand what it means for them.

David Stearns:
Thanks. That's really helpful. I feel like it's important now for us to pivot because whenever AI comes up, it feels like we're in an episode of Black Mirror and it's just this dystopian future, and we're only talking about all the negative potential implications of this. But I think it's also really important to acknowledge that AI also holds a certain amount of promise potentially to help us advance our sustainability objectives. And I'm curious if you might spend a few minutes telling us or helping us understand what are some of the ways that AI can be deployed in service of our objectives of creating a more just sustainable world? Are there any examples of where you've seen the promise of AI in an early stage where we say, "Wow, if we could really replicate this and scale it up, this could have really significant positive implications."

Lale Tekişalp:
Yes, definitely. I think AI holds real promise to advance sustainability goals, as you said. And we're seeing lots of promising applications already. For example, there's an exciting use case that we highlighted on BSR's recent Insights+ piece on AI, where AI deep learning technology is being used to discover new materials, millions of them, such as inorganic crystals that are essential to things like computer chips, batteries, and solar panels. And the best part is that these new materials have fewer environmental and social impacts compared to the conflict minerals that we currently rely on. There are lots of very impactful initiatives out there using AI for Good, AI for the Planet, etc. These are great. We should certainly use this powerful technology to advance human rights, to address the climate crisis, but we have to be cautious about the potential harms and address those potential harms because of course, benefits do not outweigh risks.

And I believe the most impactful use cases, especially on the more social applications come up when groups on the forefront of these issues use AI tools themselves. If we truly want AI to benefit everyone, we should make sure it can be leveraged by impacted communities themselves in a way that empowers them and does not extract from them. One example I want to share here is from BSR's own Tech Against Trafficking initiative, which is a coalition of companies using technology to fight human trafficking.

As part of this program, last year we worked with two anti-trafficking organizations and help them apply generative AI technologies to their work. I think this has the potential to be very impactful because these are organizations that work on the ground to address forced labor and labor abuse. And they work directly with impacted individuals and their representatives in places like Mexico and Thailand and other countries in their local languages with firsthand knowledge of the local context. So when we build and deploy AI solutions with that kind of know-how and experience, then it has real potential to help us achieve a more just and sustainable world.

David Stearns:
Thanks for that, Lale. As you know, we like to end all of our conversations with a question around what you see coming next around the corner, something that might have you excited that's coming down the line around the use of artificial intelligence, particularly in sustainable rights-respecting ways. Can you tell us anything that's on your mind that has you excited or things that we should be anticipating?

Lale Tekişalp:
Yeah. So I'm very curious to see how evolving regulations will impact the way companies manage the impacts of AI. Hopefully these regulations will have the intended outcome and push companies to think about the actual and potential impacts of the AI systems that they're using and not just be a checkbox exercise. I'm also curious to see how AI will impact the role of sustainability teams and the corporate sustainability function, both in the sense of how AI can change the day-to-day for things like ESG reporting and so forth. But also I hope to see sustainability and human rights teams take on a bigger role in shaping their company's AI strategies. Currently, the AI conversation at companies don't always include sustainability and human rights teams, and we think they need to have a seat at the table. So I'm really excited to see how that will take shape in the coming months and excited to work with our members to help them achieve that.

David Stearns:
That all sounds great, and hopefully some of our member companies will take you up on that opportunity to create those linkages between those teams internally. That will be really critical, I think. Thank you again, Lale. Very fascinating insights and I look forward to seeing more of your work on the responsible use of AI in the coming months.

Lale Tekişalp:
Thank you, David.

David Stearns:
Thanks for listening. For more in-depth insights and guidance from BSR, please check out our website at bsr.org and be sure to follow us on LinkedIn.