Charlotte Deane

Charlotte Deane grew up between South London, Jordan, and Hong Kong. She announced early that she wanted to be a scientist, initially because she could set fire to things.

She describes her work simply: ‘I do AI for drug discovery.’ More precisely, she uses computational methods to understand protein structure, protein function, and how small molecules bind to targets in the human body, making drug design faster, cheaper, and better informed.

AI entered her field gradually. Having always worked on computational, statistical, and mathematical methods, she watched machine learning tools shift from secondary tools to centre stage, a natural progression rather than a sudden career pivot. She is careful to temper the current hype: while there has been a massive jump in capabilities, at some point that jump will finish and the field will return to a more standard, ‘bumbling along’ phase of incremental progress.

The traditional drug development pipeline takes ten to fourteen years and commonly costs around two billion dollars to bring a single drug to market. Nine out of ten candidates fail somewhere along the line, after years of development, testing, and clinical trials.

The ones that do survive are not necessarily working on the diseases that most need them. Finance often dictates which treatments get prioritised because commercial pharmaceutical companies must act in the financial interests of their shareholders. She’s direct about the reality this creates: ‘A large pharmaceutical company won’t work on most of the diseases that are hurting most of the people on the planet, because they cannot recoup the cost.’

Outside of pandemic vaccines, the highest-revenue drugs of recent times have included Pfizer’s Viagra and the GLP-1 weight-loss family, driven by wealthier markets that can pay. Diseases affecting large populations in lower-income countries often remain largely untouched – not for a lack of science, but for lack of a market that can return the investment.

If computational tools can accurately predict which molecules will work before they ever hit a lab, the traditional timeline shrinks. Even a ’modest‘ shift, bringing the failure rate down from nine out of ten to seven out of ten, alters the economics considerably. It cuts years of development, saves millions of dollars in failed trials, and lowers the financial barrier to targeting diseases that disproportionately affect the world’s poorer communities.

Charlotte’s account of being a woman in science is direct. More than once early in her career, she arrived at an academic meeting only to be asked by someone to pour the coffee. It would sometimes take a few minutes for her to register the underlying assumption: that they thought she was an administrator.

But she also notices the changes. It used to be the case, she says, that every room she walked into she would probably be the only woman there. That is not so true anymore.

What she thinks has mattered most is not the numbers, but the permission to name the problem. ‘You’re allowed to say out loud that this isn’t quite right. That’s the biggest difference.‘

Support often arrived through small gestures. She remembers presenting a poster as a PhD student when Professor Dame Janet Thornton, a senior scientist who later ran the European Bioinformatics Institute, stopped to discuss her work.

A year later, at a different conference, Thornton stopped her in a corridor just to ask how her research was progressing. ‘She probably doesn’t remember doing it,’ Deane says. ‘But I do.’

For those considering a career in AI, her advice is to focus on adaptability over rigid subject choices. Quantitative foundations (maths, physics, chemistry) matter, but the willingness to pivot matters more.

Ultimately, pushing frontiers requires a massive tolerance for things going wrong. ‘Research is a career where most of what you do does not work,’ she notes. ‘But every so often there are moments – a problem solved, a result that makes sudden sense – that make all the brick walls worth running at again.’