SUMMARY
Left unchecked, artificial intelligence could entrench gender inequality. Efforts to avert this outcome may not only benefit society but also present a potential investment opportunity.
Generative AI has captured the world’s imagination.
Indeed, this technology – which can create images, text, audio and video based on word prompts – is also augmenting our imagination, as anyone who has instantly created a witty yet heartfelt poem for a colleague’s retirement party can attest.
Beyond such light-hearted applications, though, we are increasingly seeing the potential of generative AI to drive better outcomes across a whole range of fields.
Faster drug discovery, more accurate medical diagnoses, personalized learning for students, anticipating and responding to disasters, leaner supply chains, higher agricultural yields, and more efficient customer service are just a handful of the possibilities that are beginning to be realized.
But just as generative AI is already showing its capacity to make us more productive and eliminate some tedious tasks, it has also provided glimpses of a darker side.
Prompted to discuss employment avenues for various groups of people, for example, some chatbots have produced troubling answers, including offensive and misleading tropes about women and minorities.
Could generative AI make gender inequality worse?
Perpetuating stereotypes is one way that AI could end up exacerbating existing inequities.
At the heart of this issue is the fact that AI can mirror its creators’ biases.
Only 14% of tech industry leaders1 and 15% of data scientists2 – who build AI algorithms among other tasks – are women.
Male-dominated AI engineering teams could overlook women's perspectives in relation to AI policies, practices and data design. Their assumptions may reflect longstanding gender prejudices.
What’s more, the datasets that feed AI systems may reflect today’s gender-biased society.
This raises the risk of AI-powered decision making that is simultaneously discriminatory and opaque.
Intensifying use of AI by governments, companies and individuals could also reinforce economic inequities.
While AI seems likely to create some jobs (new roles to date include prompt engineers), it could potentially reduce employment in sectors where women are heavily represented.
In the US, generative AI alone could wipe out 3.7 million office support and 2 million customer service jobs by 2030, for instance.3
The risks of AI gender bias
AI is already improving some aspects of how governments, companies and the finance sector function.
Removing human error, accelerating processes, reducing costs and devising wholly new approaches are among the possibilities.
For instance, analyzing vast datasets should allow governments around the world to improve the quality of policymaking by enhancing forecasting capabilities and simulating policy changes to assess their impacts.
AI could also simplify people’s interactions with public sector bodies, making it easier to apply for state benefits or navigate healthcare systems.
However, if the AI models used to generate forecasts or engage with the public are uncritically based on past trends and behaviors, they could fail to deliver fair outcomes for women.
To take another example, companies could deploy AI to streamline hiring decisions.
But if historical hiring data showed a preference for male managers, an AI hiring tool might continue to do the same. This could create a negative feedback loop: fewer female managers could lead to ever-fewer women in senior roles, which would further bias AI against women.
Similarly, in finance, women tend to have different credit histories – due to the gender pay gap or employment breaks for childcare – that may decrease their access to some financial products. Decisions driven by AI may thus perpetuate gender disparities in access to capital.
Moreover, fewer women than men run companies. So, if an AI algorithm is trained on traditional metrics and historical lending data, it could entrench existing gender norms by unfairly favoring male business loan applicants.
In education, AI could facilitate personalized learning, effectively providing every child with a tutor who understands their learning style and their strengths and weaknesses.
In many countries, nonetheless, women are under-represented in science, engineering, technology and mathematics (STEM). AI-powered platforms risk perpetuating biases that deter women from pursuing these fields, which would ultimately widen the gender gap.
Tackling generative AI’s gender bias risks
To address the challenges in developing AI that is inclusive and beneficial for everyone, firms must make concerted efforts to train data to account for gender and other societal biases and the shortcomings of existing datasets.
AI engineers also need to remain alert to the risk that unmonitored algorithms can reinforce stereotypes.
Tech companies must actively recruit more women into AI and ensure they have better chances to progress in their careers, so that their workforce – and mindset – further reflects society. Governments also have a duty to ensure equal access to technology and digital skills training.
Implementing some of these measures will be complex, however. In particular, there is a fierce debate underway about the responsibilities of AI firms and the role of regulators and governments.
One challenge is that AI is developing faster than regulators’ ability to ensure fairness. Moreover, AI programs are proprietary technologies that developers are reluctant to open up to external scrutiny.
Another issue is that AI systems analyze personal data. If gender information is used, it can lead to privacy violations: protecting individuals’ data while ensuring fair policies is a delicate balance.
Ultimately, closing the gender gap in AI is likely to demand collaboration across governments, civil society, academia, the private sector and consumers.
Investing in AI with a gender lens
While governments and AI firms will need to reach a consensus on addressing gender bias, investors also have a crucial role to play in driving inclusive AI initiatives – and can potentially enhance their portfolios in the process. Through Gender Lens Investing (GLI), which considers gender-based factors throughout the investment process, capital can be directed toward companies and initiatives that prioritize equity and diversity in AI development.
At Citi Global Wealth, we believe the following areas offer potentially compelling inclusive AI investment opportunities:
- EdTech: Investing in AI-powered educational technology can improve access to education for women and girls, particularly in developing countries. By promoting flexible learning and inclusive curriculums, these platforms can encourage more girls to pursue STEM subjects and empower them to take advantage of emerging opportunities.
- Access to finance: AI can be leveraged to make finance more accessible to underserved communities, including women. By investing in initiatives that support women-owned businesses and facilitate financial inclusion, investors can drive economic empowerment and create opportunities for women to thrive in the digital economy.
- Healthcare: By supporting so-called FemTech firms – which use digital technology including AI to address health conditions that mainly or only impact women – investors can help improve the quality of women’s lives and narrow the gender health gap.4 FemTech could boost the global economy by $1 trillion a year by 2040.5
Inclusive practices driving impact
While it’s crucial to carefully evaluate investment opportunities with young AI-focused ventures that are unproven businesses, GLI that prioritizes equity in AI development could unleash the technology’s potential to drive positive social and economic change for women and society as a whole.
As part of our approach, we seek to identify companies with inclusive leadership teams and technology-focused funds managed by diverse portfolio managers.
Research has shown that teams which better reflect society foster innovation and provide a more balanced approach. This could potentially benefit gender-inclusive AI development.
We believe that focusing on gender equity can potentially help to diversify allocations and enhance gender diversity while you strive to make an impact through your investments.
