First AI-Designed Drugs Fall Short of Expectations: Reality vs. Hype After Years of Enthusiasm
After years of intense enthusiasm surrounding drugs designed by artificial intelligence, the pharmaceutical industry is beginning to face the reality of clinical trials. The first wave of AI-designed drugs, once celebrated as a revolutionary step in drug discovery, is encountering problems in the transition from the laboratory to actual clinical practice and has so far failed to fully meet the initial high expectations of achieving breakthrough therapeutic results.
Insilico Medicine developed Rentosertib, which became the first drug whose target and design were both discovered entirely using generative artificial intelligence. The drug successfully received an official name and progressed through clinical trials to Phase IIa for idiopathic pulmonary fibrosis. While the company reports positive results from Phase IIa trials, it is now preparing for larger pivotal studies to further test the efficacy of Rentosertib before potential regulatory approval. Although this step represents significant progress, the drug has not yet reached the market or demonstrated definitive clinical success in large patient populations. Similarly, Absci launched the first trials of its AI-designed candidate ABS-101 for inflammatory bowel disease in Phase 1 human studies in 2025. Although artificial intelligence accelerated its development and reduced costs, the primary focus remains on assessing safety and tolerability, while efficacy data have yet to be released. So far, this is a randomized, double-blind, placebo-controlled study involving approximately 40 healthy adult volunteers, designed to evaluate safety, tolerability, pharmacokinetics, pharmacodynamics, and immunogenicity.
Industry observers now point out that while artificial intelligence has significantly improved the speed and economics of early-stage drug discovery, translating these advances into clinical and commercial success remains a work in progress. As of mid-2025, no AI-designed drug has yet fully validated its promise by completing pivotal trials and obtaining regulatory approval. Although the sector continues to move forward, the transition from hype to proven clinical impact is still underway. The first generation of AI-designed drugs, including compounds such as Rentosertib and ABS-101, reached human clinical trials with remarkable speed. Despite the pace of their development, these drugs have not yet demonstrated clear clinical superiority or breakthrough efficacy in early human studies. Some have shown only modest results, while others are still gathering initial safety and tolerability data.
Rentosertib is moving forward following positive Phase IIa results, but pivotal trials are still pending and regulatory approval has not yet been secured. Specifically, in a multicenter, randomized, placebo-controlled Phase IIa study evaluating Rentosertib in patients with idiopathic pulmonary fibrosis over 12 weeks at several dose levels, the drug met its primary endpoint and showed a favorable safety and tolerability profile at all doses. Most adverse events were mild to moderate, and no serious adverse events related to Rentosertib were reported. Regarding secondary efficacy endpoints, Rentosertib demonstrated a dose-dependent improvement in lung function, specifically in forced vital capacity. The highest-dose group (60 mg once daily) recorded a mean increase in FVC of 98.4 ml from baseline, while the placebo group recorded a mean decrease of -62.3 ml.
ABS-101 entered human trials in 2025, but results regarding efficacy and clinical impact have yet to be released. In preclinical studies, ABS-101 was well tolerated at all doses during a 13-week GLP study in monkeys, with no significant safety concerns. The ongoing Phase 1 trial focuses on determining the safety and tolerability profile in humans, but no human safety data had been published as of May 2025. The muted or incremental results from these first AI-designed drugs have prompted a reassessment of expectations in the biotechnology and pharmaceutical industries. While artificial intelligence has clearly accelerated early-stage discovery and preclinical research—sometimes bringing drugs into clinical trials in half the usual time and at a fraction of the cost—demonstrating meaningful clinical benefit remains a significant challenge.
Experts note that AI can efficiently generate new chemical structures and optimize early candidates, but the complexity of human biology and disease means that translating these advances into real-world therapies is difficult and often unpredictable. The industry now recognizes that while AI is an excellent tool, drug discovery and development remain fundamentally challenging, and hype should be tempered by realistic expectations regarding timelines and outcomes. AI-designed drugs are expected to continue entering clinical trials, with hopes that improvements in data quality, model sophistication, and integration with experimental biology will lead to greater clinical impact in the future. The sector is now undergoing a recalibration, focusing on incremental progress and the long-term integration of AI with traditional scientific methods for drug development.
Artificial intelligence has transformed early drug discovery by accelerating candidate identification and reducing costs, but the first AI-designed drugs have yet to achieve significant clinical breakthroughs after years of high expectations. The field is recalibrating, focusing on incremental progress and the long-term integration of AI with traditional scientific methods for drug development. The transition from enthusiasm to proven impact in clinical practice remains ongoing, highlighting the need for realistic expectations and patience in realizing the potential of artificial intelligence in pharmaceutical research.
