I Built an AI Research Engine and It Found a Perfect Power Law in Twin Primes (2026)

Short summary

A developer built Luka, an autonomous AI research engine, and discovered a perfect power law (R² = 0.9997) in twin prime conjecture residuals using 33 verified data points across 8 orders of magnitude. Luka falsified a competing PRIT model and predicted π₂(10¹⁰) with 0.99% error. The open-source work demonstrates AI's capacity for genuine mathematical discovery beyond pattern matching.

• •Luka autonomously discovered a power law in twin prime residuals with R² > 0.9997 across 8 orders of magnitude
• •The engine falsified a competing PRIT model by two orders of magnitude and predicted values with <1.15% error
• •Open-source Python implementation suggests AI systems can make novel mathematical discoveries independently

Generated with AI, which can make mistakes.