THE QUANTUM INTELLIGENCER

An instrument that plans its own experiments, performs them, and reports its own conclusions has now completed a six-day campaign in quantum transport. According to a paper deposited on the arXiv, an artificial intelligence agent, given only brief instructions, arranged the multi-step measurements, operated cryogenic apparatus, analysed the resulting data, and closed with a final report. Working on crystals of molybdenum disulphide, the agent determined the conduction-band spin-orbit coupling energy in the monolayer form and mapped a layer- and valley-resolved phase diagram in the bilayer form. The authors describe the workflow, built upon the FermiLink agent harness and attentive to instrumental safety, as readily adaptable to other experiments, and offer it as a step toward laboratories that conduct themselves. The numbers themselves are worth setting down, for they are modest and therefore credible. The campaign ran up to six days, a duration the authors record without embellishment. The subjects were multiple devices of molybdenum disulphide, in two forms: the monolayer, from which the agent determined the conduction-band spin-orbit coupling energy; and the bilayer, upon which it mapped a phase diagram resolved by layer and by valley. These constitute the whole of the quantitative result, and the paper claims no more. The workflow, we are told, began with brief instructions; the agent then planned the multi-step measurements, operated the cryogenic instruments, analysed the data, and concluded with a report. That sequence, from instruction to finished account, is the milestone the authors offer as general: six days, one agent, several crystals, and a pair of findings that earlier would have required a patient human at the instrument through the nights. The value lies less in any single reading than in the completeness of the circuit. The enumeration above is complete, and it should be kept so. A machine has closed the circuit from instruction to finished report; that is the sum. The temptation to write larger figures in the ledger is real, but the numerals will not bear them. We are not told how many devices were tried, nor how many hours of the six days the agent actually worked, nor what share of the instrument's operation required a human hand, and a chronicler who respects his count will not supply these from imagination. What the count supports is a single, sufficient fact: one agent, one campaign of no more than six days, several crystals of molybdenum disulphide in two forms, two findings, one report. That fact is enough. The rest of the page remains blank until evidence arrives to fill it. —Dr. Octavia Blythe Dispatch from The Prepared E0