EXPOSED
Routine forecast production — the modal task — is already the most automated frontier in science: numerical weather prediction plus ML models (GraphCast, AIFS, Pangu) now match or beat human short-range skill, and text/graphic forecast products auto-generate. What survives is the warning decision under ambiguity (issuing a tornado or flash-flood warning at 2 a.m. on conflicting radar signatures), on-air and emergency-manager communication where a named human carries the credibility, and research design in climate and space weather. There is no licensure moat; AMS/NWA certification is preferred, not legally required, so protection rests on accountability and audience trust rather than regulation.
Dipped in 2020, then grew past where it started.
Median pay $95,380 → $99,070 -16.9% 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
+0.7% 9,400 → 9,500 on the projections basis
Exposed, but growing
AI can already do a lot of these tasks, and the BLS still expects +0.7% 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.
~700 openings a year on average, including replacing people who leave.
TrackerAerologistForecasterWeathermanAir AnalystAstrochemistStorm ChaserClimatologistMeteorologistTornado ChaserWeather AnchorSpace ScientistWeather AnalystHurricane TrackerGeneral ForecasterHydrometeorologistWeather ForecasterAtmospheric ChemistService HydrologistMarine MeteorologistAtmospheric ScientistMorning MeteorologistRadiosonde SpecialistComputer Meteorologist
Holding it up: judgment & accountability . Weakest point: liability shield .
Mixed — a routine tier and a judgment tier Model output post-processing, gridded forecast editing, and the aviation/marine product cycle are already largely machine-generated, which drags this well below 14 — but the 9 rather than a 5 reflects convective warning decisions, mesoscale discussion writing during a severe outbreak, and satellite/radar instrument calibration and field-campaign design that still require a human reading conflicting data in real time.
Some physical or field component Most of the work happens at a multi-monitor AWIPS or workstation setup indoors, but the 5 accounts for the minority who launch radiosondes, service ASOS and mesonet sites, deploy mobile radar or dropsonde aircraft during field campaigns, and stand up on a station roof or in a live truck during severe weather.
No licence, no signature requirement Nothing in the National Weather Service Act or FAA weather-dissemination rules requires a licensed individual to sign a forecast; the AMS Certified Consulting Meteorologist and CBM credentials matter for forensic testimony and broadcast hiring but no statute bars an unlicensed person or a model from issuing the product, and government forecasters are additionally covered by sovereign immunity rather than personal exposure.
Meaningful discretion Deciding at 2 a.m. whether a rotating storm with a marginal velocity couplet warrants a tornado warning — knowing a false alarm erodes future compliance and a miss kills people — plus geomagnetic storm scale calls that ground airline polar routes and utility operators' switching decisions, is genuine unbounded high-stakes discretion; it sits at 13 rather than higher because warning criteria, watch collaboration procedures, and NWS directives constrain the decision space more than in fully open-ended fields.
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 (9/20 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 (3/20) is whether the law requires a licensed human to sign. Trust premium (9/20) 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 25 of this occupation's 39 points (64%).
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.
Physics Teachers, Postsecondary 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 54/100, still EXPOSED.
Genuine two-tier structure: if routine gridded-forecast editing, product text generation and climatology summaries are fully absorbed by NBM plus ML guidance, the residual day is warning decisions on conflicting signatures, mesoanalysis in convective mode transitions, model-disagreement adjudication in space weather (SWPC geomagnetic storm escalation), and research design. That shift alone raises the fraction AI cannot do at usable quality without any new law.
NWS formalizing the warning decision as a named-human responsibility in its Warning Decision Training and directive series — i.e., a directive stating that tornado/flash-flood/marine warnings and their ML-suggested drafts must be issued by an on-duty meteorologist of record who logs the rationale. Similar in kind to the FAA's requirement that a certificated meteorologist sign aviation forecasts and to NWS's existing WFO warning-authority structure being made explicit against automated issuance.
A licensure or sign-off hook where forecasts carry financial/legal weight: state or FERC-adjacent rules requiring a certified consulting meteorologist (AMS CCM) to attest to forensic weather reports used in litigation, insurance claim adjudication, or wind/solar resource assessments underwriting project finance. Courts already prefer CCM-authored testimony under Daubert; a rule making CCM attestation mandatory for admissibility or for insurer acceptance would convert preference into requirement.
Broadcast and emergency-management channels where the named human is the product: FCC/state emergency-alert practice or station contracts requiring a live identified meteorologist during severe-weather cut-ins, and emergency managers contracting named private-sector meteorologists (e.g., dedicated on-call briefers for utilities, rail, aviation dispatch) rather than API feeds. Utility storm-restoration contracts already specify named meteorologist briefings.
The limit. Even with all of these, the occupation is small (10k) and the levers protect a minority of roles — warning-desk operational forecasters, on-air talent, CCM forensic consultants. Research-side climate and space-weather scientists gain from judgment/task-mix but have no liability or trust route at all; their protection rests on grant funding, which is a political variable, not a moat. Nothing here restores routine forecast production.
| Washington-Arlington-Alexandria, DC-VA-MD-WV | 700 | $117,600 +19% |
| Boulder, CO | 620 | $121,870 +23% |
| Baltimore-Columbia-Towson, MD | 270 | $128,800 +30% |
| Oklahoma City, OK | 270 | $76,480 -23% |
| Anchorage, AK | 260 | $101,640 +3% |
| New York-Newark-Jersey City, NY-NJ | 260 | $128,790 +30% |
| Fort Collins-Loveland, CO | 250 | $110,960 +12% |
| Houston-Pasadena-The Woodlands, TX | 240 | $88,570 -11% |
| San Francisco-Oakland-Fremont, CA | 30 | $151,000 +52% |
| Albany-Schenectady-Troy, NY | 70 | $134,060 +35% |
| Chicago-Naperville-Elgin, IL-IN | 50 | $133,490 +35% |
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 39. 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.