Unlock Editor’s Digest for free
FT editor Roula Khalaf selects her favorite stories in this weekly newsletter.
The AI hype cycle depends on eye-catching, record-breaking ads. In April, San Francisco-based startup Xaira did just that: announcing which had raised $1 billion in one of the biggest launches in the history of biotechnology.
Xaira says drug development is primed for an AI revolution. You are not alone. Demis Hassabis, co-founder of Google DeepMind, famous for solving the 50-year scientific challenge of predicting protein shape, argues that biology could be “perfect” for AI, since it is fundamentally an information processing system. He runs Isomorphic Labs, Alphabet’s AI drugs arm that has agreed to partnerships worth up to $3 billion with Eli Lilly and Novartis. This belongs to me reduce by half the drug discovery stage just two years away.
An increasing number of AI-derived compounds are being developed. The World Health Organization has identified at least 73, although none are yet approved for use in humans. Some companies are reaching out. Insilico Medicine, which recently filed for an initial public offering in Hong Kong, was the first to include an AI-designed drug in Phase II clinical trials.
But AI still does not replace the experimentation that underpins the understanding of a disease. The sector has already experienced problems. On the day of Xaira’s launch, BenevolentAI announced major layoffs. The London-based company set out to unite human and machine intelligence, but its shares have lost 94 percent of their value since it went public at a valuation of €1.5 billion in December 2021 through a merger with a company special purpose acquisitions.
Developing innovative new medicines is expensive and inefficient. The pharmaceutical industry has no shortage of funding or motivation when it comes to improving drug discovery success rates using AI. Around 200 biotech companies that prioritize AI have raised more than $18 billion in the decade to 2023, according to the consulting firm BCG. Both AI usage and success rates vary.
The use of computer science in drug design is nothing new, dating back to the 1970s. The insights are only as good as the data used to train models. Predicting the toxicity of drug candidates is hampered by the paucity or relevance of publicly available information. There is a lot of data on profitable and highly sought after fields of research, such as cancer. There are fewer in relatively neglected areas such as mental health or infectious diseases.
AI is not a magic solution to these problems. Data gaps can be filled through experimentation, but it takes time and a lot of money.