Opus 5.5 agents discover two room-temperature magnetic semiconductor candidates
Points and comments are a snapshot, not live.
AI agents predicted two room-temperature antiferromagnetic semiconductors for spintronics.
A blog by an AI researcher describes using LLM-driven agents with DFT simulations (PBE+U, HSE06) to find two Luttinger-compensated (LC) magnets: YBaMnFeO₅ (designed, but likely hard to synthesize due to checkerboard ordering) and KV[Cr(CN)₆] (first made in 1999, already stable above room temperature at 376 K). Both show zero net magnetism, semiconductor band gaps (~2.1-2.35 eV), and large spin-sorting windows (1.0-2.6 eV) far above thermal fluctuations. The author notes KV[Cr(CN)₆] was hiding in plain sight.
The next step is to remake KV[Cr(CN)₆] and measure its spin sorting directly.
What commenters are saying
Skepticism dominates: many commenters misread the title as "room-temperature superconductor" (due to LK-99) and dismiss the theoretical predictions without experimental verification. A common refutation is that DFT simulations are standard graduate-level work, not breakthrough science. Two camps form: those who see LLM-driven discovery as a powerful tool, and those who argue the expensive part is synthesis and measurement, not computation. A few push back, noting that ideas worth testing are valuable even if unverified.