AI Safety

Why clinical trials are broken & how to fix them: a reading list

Drug development costs have increased 80x since the 1950s, now requiring ~$1B per approved drug.

Deep Dive

A growing movement of researchers and patient advocates, organized under the banner 'Clinical Trial Abundance,' is diagnosing why drug development has become prohibitively expensive and slow. Their central thesis, popularized by Jack Scannell's 2012 paper, is 'Eroom's Law'—the observation that the cost of developing a new drug has increased by roughly 80x since the 1950s, now requiring about $1 billion per approved medication when accounting for failures. This trend has led to fewer drugs being invented, a systematic avoidance of ambitious but risky research areas (like treatments for ME/CFS or Long COVID), and ultimately, patients paying the price through lack of innovation.

A curated reading list, compiled by Siebe on LessWrong, serves as an entry point to the movement's ideas. Key articles include work by Alex Telford on the historical shift from small, quick trials to massive preclinical projects, and a 2025 piece by movement leader Ruxandra Teslo and Scannell titled 'To Get More Effective Drugs, We Need More Human Trials.' This article explicitly debunks two common myths: that a purely libertarian approach to drug approval will work, and that advanced AI will automatically solve the problem. The movement argues that rigorous human testing remains irreplaceable, and the current system's risk aversion and regulatory caution are major bottlenecks that need systemic reform to increase the sheer volume of clinical trials conducted.

Key Points
  • Identifies 'Eroom's Law': Drug development costs have increased ~80x since the 1950s, now ~$1B per approved drug.
  • Debunks silver bullets: Argues neither AI nor simple deregulation can fix the system without more human trials.
  • Highlights systemic failure: Current system avoids risky bets (like for Long COVID) due to cost and regulatory caution.

Why It Matters

Fixing clinical trial inefficiency could accelerate treatments for chronic and neglected diseases, impacting millions of patients.