Vivodyne says AI drug discovery needs better human tissue data

In short: Vivodyne says AI is not close to curing cancer because it lacks good data from human biology, and the company built robotic labs to create that data.

What happened

A biotech startup called Vivodyne told TechCrunch that much of today’s AI drug discovery work is built on the wrong kind of information. The company says AI models are often trained on animal tests, or on simplified lab studies that look at single cells or proteins instead of living human tissue.

Vivodyne’s answer is a system it calls HIVE. These are modular robotic labs that can grow about 20 types of human tissue and then automatically add drugs and track what happens over time. Vivodyne says this creates “causal” data, meaning it shows what changed and why, more like watching a full cooking video instead of seeing only a photo of the finished meal.

CEO Andrei Georgescu argues that without this kind of human-focused data, AI systems may mainly find drugs that work in animals but fail in people. That matches a broader problem in medicine: TechCrunch reports that around 90% of drugs that look promising in animal tests still do not get approved for humans.

TechCrunch also reports that Vivodyne opened what it calls the world’s largest “human data center” near San Francisco. The company has raised just under $80 million in two funding rounds led by Khosla Ventures, and says it is working with multiple large pharmaceutical companies, though it did not name them.

Why it matters

Many people hear claims that AI will quickly cure cancer, but real progress depends on reliable human testing data. If Vivodyne’s approach works, drugmakers could learn earlier which ideas are likely to fail, which may save time, money, and unnecessary trials.

Source: TechCrunch AI