EXPOSED
Astronomy is largely screen work — pipeline code, photometry and spectral fitting, catalog cross-matching, literature review, and manuscript drafting — and AI already does credible versions of most of it, with ML classifiers outperforming humans on transient and morphology sorting. What resists is deciding which questions are worth years of telescope time, designing and commissioning instruments, defending survey systematics, and owning the claim when a result is extraordinary. There is no licensure and no signature requirement; the field's real constraint is that only ~2,000 US positions exist and they are gated by grants and faculty lines, not by automation.
Roughly flat across the period, with year-to-year wobble.
Median pay $114,590 → $128,820 -10.1% in real terms
This line is counted by the Bureau of Labor Statistics — the one figure on this page that isn't a judgement of ours. Headcount moves on demand, offshoring, demographics and the business cycle, and automation is one term among several, often not the loudest.
So a falling line is not evidence that AI did it, and a rising one is not evidence that it won't. Both happen in this register: some occupations resist automation and shrink anyway, others are highly automatable and keep growing. The marked year is 2020.
BLS projection, 2024–2034
+2.2%
Percentage only. The projection counts a different population from the 2,120 above — it includes self-employed workers, which for this occupation is most of them, so the two headcounts are not comparable.
Exposed, but growing
AI can already do a lot of these tasks, and the BLS still expects +2.2% more of these jobs by 2034. Demand for the output is growing faster than the work is being automated away — the mechanism BLS gives for software developers, and the combination people most often misread as an error.
Different clocks. The score is what current AI could do to this work today. The projection is how many of these jobs will exist in 2034. Everything between the two — how fast employers actually adopt, whether demand grows in the meantime — is why they can point opposite ways without either being wrong.
~100 openings a year on average, including replacing people who leave.
AstronomerCosmologistAstrophysicistRadio AstronomerSolar AstronomerOptical AstronomerStellar AstronomerGalactic AstronomerInstitute ScientistPlanetary AstronomerTheoretical AstronomerResearch AstrophysicistExtragalactic AstronomerHigh-Energy AstrophysicistAstronomy Outreach CoordinatorPostdoctoral Scholar (Postdoc Scholar)Postdoc Scientist (Postdoctoral Scientist)Postdoctoral Associate (Postdoc Associate)Postdoctoral Research Associate (Postdoc Research Associate)
Holding it up: judgment & accountability . Weakest point: liability shield .
Mixed — a routine tier and a judgment tier Reduction pipelines, aperture photometry, redshift fitting and cross-matching against Gaia/SDSS catalogs are already scripted and increasingly ML-driven, but writing a competitive TAC or NASA ADAP proposal, choosing the observing strategy that survives moonlight and weather constraints, and diagnosing an unexpected instrumental systematic in the data still take a person who knows the hardware — that mix of automatable analysis over irreplaceable framing is why this lands at 10 rather than 5 or 15.
Some physical or field component Most astronomers work from an office and a queue-scheduled interface, with the physical component limited to observing runs at Kitt Peak or Mauna Kea, cryostat and detector work for instrument builders, and occasional lab alignment — real but episodic and confined to controlled domes, which is why it sits at 5 rather than the 12+ of someone whose week is spent on a mountain.
No licence, no signature requirement There is no state licence, no board certification, and no stamped deliverable in astronomy; the only formal gates are a PhD and peer review, and a retraction damages a reputation rather than triggering legal exposure — hence 1 rather than 0, since institutional review boards and export-control rules touch some instrument and defense-adjacent work.
Meaningful discretion When you claim a planet in the habitable zone, a 5-sigma detection at the edge of your noise floor, or a cosmological parameter that disagrees with Planck, you personally decide whether the systematics are understood well enough to publish, and that call is unfalsifiable in the moment and career-defining afterward — 13 rather than higher only because no one's safety or money rides on the answer.
The verdict above describes this occupation as a whole. Almost nobody does the typical version of a job — tick what's actually in your week and see how your own mix sits.
Your task mix speaks to task resistance (10/20 here) — how much of the day's work current AI already does. That is the dimension the boxes above are about.
It cannot move the other three. Liability shield (1/20) is whether the law requires a licensed human to sign. Trust premium (8/20) is whether buyers specifically pay for a person. Judgment and accountability (13/20) is whether the role exists to own consequential calls. Those are facts about the occupation's standing, not about which tasks are in your week — a paralegal who does only trial exhibits still holds no licence. Together they are 22 of this occupation's 37 points (59%).
Embodiment (5/20) is also a property of the work rather than the worker, but we don't tag individual tasks as physical or not, so the picker can't tell you anything about it. That's a limit of this tool, not a claim.
Did we get the list right? Tell us what's missing — the tasks are written from the outside, and you're reading this from the inside.
Aerospace Engineers EXPOSED
Engineers, All Other EXPOSED
Nuclear Engineers EXPOSED
The moves above are yours to make. This is the other half: what would have to change in the world for the occupation itself to score higher. None of it is in any one person's gift, but it is where the floor actually comes from. Scores here are not a one-way ratchet. Only two of the five dimensions — task resistance and embodiment — track what machines can do. The other three track law, what buyers will pay for, and who is answerable, and those move in both directions, often in response to the same pressure AI creates. If every lever below landed, this occupation would score around 49/100, still EXPOSED.
Task-mix shift is genuine here: as ML pipelines absorb photometry, transient triage and catalog cross-matching (already the norm on ZTF/Rubin brokers), the residual job becomes instrument commissioning, systematics forensics on survey data, and TAC-level science-case design. If Rubin/Roman-era systematics work becomes the dominant funded activity — as LSST DESC's systematics working groups suggest — the day's work is the tier AI cannot close.
If journals and collaborations formalize a named-human 'analysis guarantor' for extraordinary claims — the way AAS journals already require author-contribution statements and some collaborations require internal review before publication, and as astro-ph adopts arXiv's policy of rejecting undisclosed AI-generated submissions — the person who signs off on a detection claim owns it explicitly.
If a larger share of positions shift toward instrument build, integration and on-mountain commissioning — adaptive-optics alignment, cryostat and detector work, Rubin/ELT/Roman hardware — the physical, unpredictable-environment fraction of the occupation rises. This is a real reallocation of NSF/NASA money toward facilities, not a new capability.
Time-allocation committees and instrument-review boards (NASA TAC, NSF/NOIRLab, ESO OPC) adopting explicit rules that AI-drafted proposals must have a named human PI accountable for the science case, and barring AI from panel review, keeps the consequential allocation calls with identified humans.
The limit. No route to a liability shield: there is no astronomy licensure, no statutory signature, and no injured party with standing to sue over a misfit spectrum. Trust premium is also close to capped — buyers are grant panels and other astronomers, not consumers who would pay extra for a human. The binding constraint on this occupation was never automation; it is the ~2,000 grant- and faculty-gated positions, so headcount can fall on flat NSF budgets regardless of any score here.
| Washington-Arlington-Alexandria, DC-VA-MD-WV | 520 | $156,750 +22% |
| Boston-Cambridge-Newton, MA-NH | 180 | — |
| Washington-Arlington-Alexandria, DC-VA-MD-WV | 520 | $156,750 +22% |
We have no reported case of a named organisation automating this occupation. Not one deployment, not one announcement.
That is worth saying out loud next to a score of 37. The verdict above is about what the work exposes — what current AI could do to these tasks. It is not a claim that anyone has done it. For this occupation those two things have come apart completely: the capability argument is on this page, and the evidence column is empty.
Has AI actually changed your work? One tap, anonymous, and the running tally is public. Nothing else is asked of you.
Rather than check back: get the digest and we'll tell you what changed — or watch a single occupation from its own page.