rationale:

The strict bar for AI attribution. At 42%, "None by 2035" is the most likely outcome, driven by the rigid criteria for what counts as a resolution. A general-purpose AI system—not just a specialist classifier—must produce the central idea, and the astronomical community must explicitly credit it AI uncovers the universe's 'settings' with unprecedented ... The role of AI agents in cosmology - Indico - CERN. While agentic AI has proven capable of automating the "mechanical middle" of research, such as a crew of 30 agents running a full supernova analysis About 30 LLM agents did a cosmology project with no human ... or AI vetting 373,646 variable stars, astronomy remains culturally and practically geared toward crediting large human-led collaborations. Consensus moves slowly and relies on multi-dataset verification, meaning AI may do the heavy analysis while human collaborations retain the scientific credit.

The capability trajectory vs. astronomical reality. AI capabilities are advancing remarkably fast, with FrontierMath Tier 4 completion jumping to 100% by September 2026 and agent-hours scaling exponentially. Mathematics, where results can be formally verified (e.g., in Lean), allows AI to make rapid, definitive breakthroughs. Astronomy lacks this formal verification mechanism. Nevertheless, rapid anticipated progress in theoretical reasoning and data mining makes it likelier that AI tackles a major cosmological problem, supporting a combined 58% probability of an AI-credited resolution by 2035 . Furthermore, resolving major questions often hinges on physical data that does not yet exist. Future observations from DESI, Euclid, Rubin, and Roman will be critical for testing cosmological theories Is the Hubble Tension Signaling New Physics? Dark Energy and Cosmology Science Updates. Even if AI rapidly processes this future data, a straightforward empirical measurement that settles a debate would likely not be credited as an AI-originated central idea.

Hubble tension and fast radio bursts. If an AI-credited resolution occurs, the Hubble tension (16%) is the leading candidate. It is the most prominent crisis in cosmology, sitting at a 5 to 7.1σ discrepancy The Hubble tension: A decade review, and is actively targeted by AI systems hunting for systematic errors across datasets. An AI-discovered systematic in the distance ladder or a novel cross-dataset theoretical framework that achieves community consensus is a highly plausible path. The fast-radio-burst (FRB) emission mechanism (12%) is the next strongest candidate. It features rich public datasets (like CHIME) and unresolved theoretical physics—competing magnetospheric and shock explanations Fast radio burst The Second CHIME/FRB Catalog of Fast Radio Bursts—making it an ideal target for a general agent to supply a purely theoretical breakthrough without waiting for next-generation observatories.

Data-constrained and detection-limited questions. The remaining questions are largely constrained by planned surveys or the absolute need for physical detections. Whether dark energy evolves (10%) and the S8 tension (8%) are deeply data-bottlenecked. While KiDS-Legacy and DES Year 6 differ KiDS [2601.14559] Dark Energy Survey Year 6 Results, humans may settle S8 with upcoming survey data before an AI can claim a unique central idea. The cosmological lithium problem (6%) offers a theoretical opportunity regarding stellar depletion The ups and downs of inferred cosmological lithium, but establishing a population-wide explanation is historically slow. Finally, identifying dark matter's particle nature (3%) and the origin of ultra-high-energy cosmic rays (3%) remain firmly tied to experimental measurements The LZ Dark Matter Experiment | The status and science of the ... 30. Cosmic Rays, leaving almost no room for a purely AI-generated idea to take central credit before 2035.