5 Insights on Generative AI, Social Impact and the Role of Business

Article co-authored with Jane Nelson, Director, Corporate Responsibility Initiative, Harvard Kennedy School.

Generative Artificial Intelligence (AI) is triggering transformations that were hard to envisage even a year ago, with some predicting it could be as transformative as the Industrial Revolution. Generative AI, which includes systems such as OpenAI’s ChatGPT, Microsoft’s Bing Chat, Google’s Bard, and Anthropic’s Claude, can generate new content or predictions based on vast amounts of data and computing power and deliver this to users in eerily human-like responses.

Somewhere between the unbridled optimism and the existential angst associated with the sheer speed, scale and uncertain impact of this rapidly evolving and transformative technology, is an immediate and urgent reality. If other systemic shocks and transformations are anything to go by—such as the impacts of the global pandemic or climate change and the drive to decarbonisation—the gains and risks of AI will reflect deep-seated inequities such as income, gender, race and geography.

In a new paper, we highlight some potential opportunities and risks that generative AI brings for the lives, livelihoods and access to learning of vulnerable people and communities. We offer a practical framework* to guide business action in mitigating these risks and realising the social potential of generative AI: leveraging core business activities and value chains, corporate philanthropy and social investment, and engagement in policy dialogue and advocacy.

Our paper is intended to spark a conversation across our community and beyond on what is a rapidly evolving area. Here we offer five insights to frame the discussion. We would love to hear your perspectives, via this two-minute survey, on the opportunities and risks and how we can work together.

1. Engage in constructive dialogue to better understand the risks and opportunities?

These so-called “large language models” (LLMs) have delivered some remarkable results that have surprised even their creators. GPT-4 scored in the 90th percentile in the Uniform Bar Exam, an exam in the US for those wanting to become lawyers. Meanwhile, Google’s AI taught itself Bengali, even though it was never trained to speak it.

At the same time, generative AI is becoming more accessible for non-experts. Many organisations, from the educational NGO Khan Academy to the French supermarket chain Carrefour, are embedding generative AI to improve the value they can offer to their customers, bringing generative AI into the mainstream. While AI has been around for decades, generative AI and its integration into a suite of new applications mean it will become ever more pervasive in the coming months and years.

The rapid growth and fast-evolving capabilities of generative AI have led to growing concern about its social, economic and political implications. Policymakers worldwide are scrambling to understand how to regulate and control AI risks while incentivising corporate innovation and investment or at least not stifling them. Companies, NGOs and entrepreneurs active in the social impact and development community need to understand better the social impacts of developing and using AI and engage constructively in dialogue with governments and each other.

2. Leverage the transformative potential of generative AI for positive social impact

Generative AI holds immense potential for positively impacting people’s lives, livelihoods and access to learning. One example is the promise AI offers in terms of transforming health outcomes: accelerating medical research and drug discovery, assisting clinical decision-making to enable early detection and diagnosis, boosting health capacity in the face of a vast shortage of healthcare workers, and increasing access to health information and services in a way that is also personalised.

Generative AI offers an important way to tackle poverty through its impact on productivity and growth. Goldman Sachs predicts the new wave of AI could “drive a 7% (or almost $7 trillion) increase in global GDP and lift productivity growth by 1.5 percentage points over a 10-year period”, though, of course, equity—including access to new jobs in an AI-driven economy—will shape the impact on poverty. Social impact and development organisations can also benefit from AI efficiency gains, enabling them to do more with their time and resources.

Generative AI is already transforming access to education and personalising this access. For people around the world, whether students, employees, entrepreneurs, or smallholder farmers, AI chatbots could provide access to learning tailored to their individual needs. The growing prevalence of smartphones will amplify this, and the ability of generative AI to understand and deliver responses in multiple languages will further open access.

3. Address and mitigate the substantial social risks posed by generative AI

While the potential benefits of generative AI are immense, it also brings significant risks and challenges that must be carefully navigated. Existing racial and gender biases in AI systems and their algorithms have been well documented, and generative AI is likely to amplify this. Last year, the American Civil Liberties Union flagged the risk of “AI and algorithmic tools that exacerbate racial biases” being used in clinical decision-making. While this reflects the biases of the humans who programme the AI and those who use it to make decisions – it also reflects a lack of diversity in the training data AI uses to generate outputs.

Another clear impact is the real threat of job losses that comes with productivity gains. The same Goldman Sachs research, cited earlier, suggests that generative AI “could expose the equivalent of 300 million full-time jobs to automation”. This does not mean all these jobs will be lost; in many cases, AI will displace tasks rather than people. Nevertheless, job losses are coming, and we know from experience the risks will likely reflect deep-seated inequities. One study suggests AI will “disproportionately replace jobs typically held by women”. Meanwhile, access to the new jobs that will be generated will not necessarily be equal. According to the World Economic Forum, only 22% of AI professionals are women.

A priority from a social impact perspective is to understand the barriers people face in accessing AI's opportunities. One obvious barrier in terms of access to learning is digital exclusion. According to the International Telecommunication Union, 2.9 billion people have never used the Internet, with 96% living in developing countries. In the UK, which recently announced its ambition to become a global AI superpower, around 10 million people “lack the very basic foundational skills needed for our digital world”.

4. Embrace the crucial role of business in harnessing the positive social impact and mitigating the risks of generative AI

Governments must take the lead in understanding and regulating AI's rapid evolution and systemic impacts, both within their own countries and coordinating globally. At the same time, businesses have a critical role to play.?Through their core business operations, philanthropy, and policy engagement, businesses can help to ensure that the benefits of generative AI are more widely shared and that risks are effectively managed with respect to people’s lives, livelihoods and access to learning.

Through their core business, companies can, for instance, harness AI to improve workplace health and safety. They can conduct AI social impact and human rights assessments to identify the potential opportunities and challenges of how AI might impact workers, suppliers, customers and communities across the value chain and ensure that the use of AI is consistent with existing human rights and Diversity, Equity and Inclusion frameworks and commitments, including where there are job losses.?They can provide employees with inclusive opportunities for training to get the skills they need to thrive in an AI-enabled workplace.

Through their philanthropic giving, companies can fund AI projects that aim to improve health and educational outcomes in vulnerable communities. They can fund NGOs to incorporate AI into their work and build reliable datasets that better reflect the communities they serve. They can also fund research into AI for social good and on the social impact of AI.

And through their policy engagement, companies can, among other things, advocate for policies and regulatory frameworks that ensure the safe and ethical use of AI and participate in constructive stakeholder discussions on these topics.?They can engage in policy dialogue to support a just transition in the face of AI-driven labour market shifts, ensuring gains are more fairly distributed, labour rights are respected, and social protection is provided for those negatively impacted.

5. Foster collaborative partnerships to shape a future that fosters equity and resilience

The complexity and systemic nature of AI’s impacts requires businesses to work together with each other and with civil society and government across silos and sectors to understand AI's risks and opportunities better and share good practices on AI adoption and integration as these emerge. Importantly, this should include the voices and experiences of vulnerable people and communities proximate to the challenges that must be addressed.

There is also an urgent need to work together to craft smart regulatory frameworks within countries and ensure global coordination. These are especially needed to put safeguards in place to protect people against the risks of AI, as well as policies to ensure that the benefits are more fairly shared and that people negatively impacted by AI are supported through training and social protection.

As we move forward, it is our collective responsibility to ensure that the benefits of AI are accessible to all and that its potential risks are understood and mitigated. The future of AI will be determined by the decisions made by today’s policymakers and leaders in business and civil society. It is up to everyone to work collectively to shape the spread and use of AI in a way that serves the best interests of society, including the most vulnerable people and communities.

>> DOWNLOAD THE FULL PAPER HERE

>> SHARE YOUR PERSPECTIVES VIA THIS TWO-MINUTE SURVEY

* This framework used in our paper "Generative AI and Social Impact: The Role of Business" was originally developed by Business Fights Poverty and Jane Nelson, Director of the Corporate Responsibility Initiative at Harvard Kennedy School, as part of our joint work on business and COVID-19. It builds on a framework developed by Jane Nelson on how businesses can engage in sustainable development issues.