AI Drug Discovery Hype vs. Reality, Fusion Energy Timeline Skepticism, Social Media Mental Health Claims

Insilico Medicine's AI-discovered drug ISM001 entered Phase 3 clinical trials for idiopathic pulmonary fibrosis, generating headlines about AI revolutionizing drug discovery. But here's the context most coverage omits: Phase 3 trial success rates are historically 50-60%, regardless of how the compound was discovered. AI primarily accelerated the discovery phase (2 years vs. typical 4-5), but discovery is only 10% of total drug development cost and timeline. The real bottleneck — clinical trials, manufacturing, and regulatory approval — remains unchanged. Furthermore, of 20+ AI-discovered compounds that entered Phase 1 trials since 2023, only 3 have progressed to Phase 3. That's a 15% advancement rate, roughly in line with traditional discovery methods.

Commonwealth Fusion Systems announced their SPARC reactor achieved net energy gain of Q=1.5 in a sustained 30-second plasma burn — a genuine scientific milestone. Headlines proclaimed 'fusion energy is here.' But the path from Q=1.5 in a lab to commercial power generation requires: sustained Q>10 (7x improvement), continuous operation (not 30-second bursts), a tritium breeding blanket (none exists at scale), turbine integration, grid connection, and regulatory approval. The most optimistic timeline for a commercial fusion plant: 2040. The realistic timeline: 2045-2050. For context, ITER — the international fusion project — has been under construction since 2013, is $25B over budget, and isn't expected to achieve first plasma until 2034.

A widely-cited study claimed that Australia's under-16 social media ban led to a '23% reduction in teen anxiety symptoms' within 3 months. Major outlets ran with the headline uncritically. However, the study relied on self-reported survey data (notoriously unreliable for mental health), had no control group, measured only 3 months of data (novelty effects fade), and was conducted during school holiday periods when anxiety naturally decreases. A Lancet Psychiatry editorial noted that the most rigorous meta-analyses on social media and mental health show 'small and inconsistent' effect sizes, suggesting the relationship is far more complex than 'social media bad.' The policy may still be justified on other grounds, but this study doesn't prove what headlines claim.