The bottleneck for AI in drug development isn't the sophistication of the models but the absence of large-scale, high-quality biological data sets. Without comprehensive data on how drugs interact within complex human systems, even the best AI models cannot make accurate predictions.
Related Insights
Powerful AI models for biology exist, but the industry lacks a breakthrough user interface—a "ChatGPT for science"—that makes them accessible, trustworthy, and integrated into wet lab scientists' workflows. This adoption and translation problem is the biggest hurdle, not the raw capability of the AI models themselves.
The primary barrier to deploying AI agents at scale isn't the models but poor data infrastructure. The vast majority of organizations have immature data systems—uncatalogued, siloed, or outdated—making them unprepared for advanced AI and setting them up for failure.
The next leap in biotech moves beyond applying AI to existing data. CZI pioneers a model where 'frontier biology' and 'frontier AI' are developed in tandem. Experiments are now designed specifically to generate novel data that will ground and improve future AI models, creating a virtuous feedback loop.
Professor Collins’ team successfully trained a model on just 2,500 compounds to find novel antibiotics, despite AI experts dismissing the dataset as insufficient. This highlights the power of cleverly applying specialized AI on modest datasets, challenging the dominant "big data" narrative.
In a direct comparison, a medicinal chemist was better than an AI model at evaluating the synthesizability of 30,000 compounds. The chemist's intuitive, "liability-spotting" approach highlights the continued value of expert human judgment and the need for human-in-the-loop AI systems.
While AI can accelerate the ideation phase of drug discovery, the primary bottleneck remains the slow, expensive, and human-dependent clinical trial process. We are already "drowning in good ideas," so generating more with AI doesn't solve the fundamental constraint of testing them.
For years, access to compute was the primary bottleneck in AI development. Now, as public web data is largely exhausted, the limiting factor is access to high-quality, proprietary data from enterprises and human experts. This shifts the focus from building massive infrastructure to forming data partnerships and expertise.
The effectiveness of an AI system isn't solely dependent on the model's sophistication. It's a collaboration between high-quality training data, the model itself, and the contextual understanding of how to apply both to solve a real-world problem. Neglecting data or context leads to poor outcomes.
The primary reason multi-million dollar AI initiatives stall or fail is not the sophistication of the models, but the underlying data layer. Traditional data infrastructure creates delays in moving and duplicating information, preventing the real-time, comprehensive data access required for AI to deliver business value. The focus on algorithms misses this foundational roadblock.
A critical weakness of current AI models is their inefficient learning process. They require exponentially more experience—sometimes 100,000 times more data than a human encounters in a lifetime—to acquire their skills. This highlights a key difference from human cognition and a major hurdle for developing more advanced, human-like AI.
