The world may be in the middle of an AI arms race – but new research has suggested that female-led startups are missing out.
A report from the Alan Turing Institute’s Women in Data Science and AI team revealed that while global investment in AI is booming, the vast majority of funding is flowing into male-dominated companies, with 77% of venture capital investments in the UK last year going to startups without any female founders. This occurs even as the number of female-founded AI companies that are launching, rises.
Analyzing data from Pitchbook, researchers found that despite VC firms investing £35 billion [$42.5 billion] in UK-based AI software firms over the past decade, less than half a percent of that funding went to companies with all-female founder teams.
These women-led startups raised six times less on average than those founded by men over the last decade, the data showed.
The report suggests that female-founded companies are missing out on the AI investment bonanza that began with the launch of ChatGPT last year.
And researchers warned that the gender disparity risks exacerbating the well-documented issues many AI systems have with bias and discrimination.
“When we talk about designing responsible and safe AI, it’s really important that we consider what’s happening in the whole of the AI ecosystem,” Turing Institute research fellow and project co-lead Dr Erin Young told Insider.
“That obviously includes things like AI design, the kind of data being used to train these models – but it also includes the funding channels.
“We need to think about the types of AI technologies that are being funded and how these AI technologies are being used, who they might benefit, and who they might harm,” she added.
Aligned AI CEO Rebecca Gorman told Insider that female founders are often subject to more scrutiny than their male counterparts – something that is now hurting VC firms as global economic conditions take a turn for the worse.
“Funding a founder because he met a stereotype was a successful strategy during the zero interest period, as the most important quality was whether he would raise a larger round of funding 12-18 months down the road,” she said.
“Funds that continue to follow this strategy will learn too late that they’ve been selecting the wrong founder and startup characteristics for a post-zero interest rates world,” she added.
Bias against marginalized groups has long been an issue with AI, with generative models having a tendency to inherit prejudices from the vast datasets they are trained on.
OpenAI’s DALL-E 2 image generator was found to overwhelmingly produce images of white men in response to prompts for “CEO” and “director” upon its release last year, and Amazon was forced to scrap an AI recruiting tool in 2017 after realizing that it discriminated against women.
A debate is now raging over how to prevent the proliferation of biased models and products, with California close to passing a law that would require VC companies to disclose the race and gender of the founders of the start-ups they fund.
Angela Hoover, co-founder and CEO of San Francisco-based AI search engine company Andi, told Insider that the lack of diversity in the AI and venture capital industries is a concern.
“I do worry about bias and discrimination in AI models,” she said. “One of our biggest priorities is ensuring AI models are fair and unbiased,” she said.
“While AI is heavily male-dominated and we still see a big gender disparity at AI events in San Francisco, we’re hopeful this will change as more women become involved in AI companies. The more diversity we have in AI, the better,” she added.
Dr Young told Insider that VC firms need to improve their recruitment policies, foster more inclusive cultures, and monitor their investment strategies to mitigate bias to ensure responsible AI development.
“This recent explosion in investment in generative AI has made the need for women and marginalized groups to have an equal place in tech entrepreneurship and the VC ecosystem much more urgent,” she said.
“It is crucial that we encourage inclusion and equity in these fields to promote responsible AI design,” she added.