EXPOSED
This catch-all category spans bench researchers, field biologists, and agency/industry science staff; the modal worker splits time between hands-on protocols (culturing, sampling, assays, specimen handling) and screen work (literature synthesis, statistical analysis, manuscript and grant drafting, protocol writing) — and the screen half is exactly what LLMs now do at usable quality. Wet-lab and field execution stays human because lab automation is capital-intensive and biological samples are messy, but no license protects the role and nobody hires a biologist for the relationship. Experimental design and interpretation under genuine uncertainty are the durable core; routine analysis and writing are not.
Headcount grew steadily across the period.
Median pay $82,220 → $98,920 -3.8% 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
+1.2% 63,700 → 64,500 on the projections basis
Exposed, but growing
AI can already do a lot of these tasks, and the BLS still expects +1.2% 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.
~4,800 openings a year on average, including replacing people who leave.
BotanistBiologistScientistAlgologistBioanalystBryologistEtiologistGeneticistMycologistResearcherTaxonomistBioassayistOsteologistData AnalystEmbryologistEntomologistMorphologistNematologistPhysiologistPaleobotanistAstrobiologistCell BiologistNeurobiologistOlericulturist
The BLS uses Biological Scientists, All Other for work that doesn't fit any named occupation, so it covers roles that have little in common with each other. Two consequences worth knowing before you read anything below:
Holding it up: embodiment . Weakest point: liability shield .
Mixed — a routine tier and a judgment tier At 11 the split is roughly even: pipetting a qPCR plate, dissecting a specimen, or running a transect through wetland cover has no software substitute, but the literature review, sequence alignment pipelines, R/Python analysis scripts, methods sections and grant boilerplate that fill the other half of the week are now first-draft work for a model, which is why this sits below the 14 line rather than at it.
Some physical or field component A 12 reflects that the bench work is real but mostly indoors and controlled — BSL-2 hoods, incubators, cryostorage, microscopy — with field sampling, animal handling, or greenhouse and mesocosm work adding uncontrolled conditions for a substantial minority of the category rather than all of it.
No licence, no signature requirement There is no biologist's license: you can run a lab bench with a BS and no board exam, and where oversight exists it flows through the institution — IRB, IACUC, IBC, or a licensed veterinarian or PI who signs the protocol — so nobody can be barred from practice, only from a grant.
Meaningful discretion At 12 you own the calls that shape the science — which controls a design actually needs, whether a contaminated run gets discarded or reported, how to interpret an ambiguous population trend or an off-target effect — but those calls are checked by PIs, reviewers, and replication before anyone's health or money is on the line, which is what separates this from the 14+ band.
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 (11/20 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 (2/20) is whether the law requires a licensed human to sign. Trust premium (6/20) is whether buyers specifically pay for a person. Judgment and accountability (12/20) 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 20 of this occupation's 43 points (47%).
Embodiment (12/20) 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.
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 56/100, still EXPOSED.
FDA/EPA data-integrity expectations extended to computational and AI-assisted work — e.g. FDA's January 2025 draft guidance on AI in regulatory decision-making for drugs and biologics requires credibility assessment and a human accountable for model context of use; if study directors under GLP (21 CFR 58) are formally required to own AI-derived conclusions in submissions, the interpretive call becomes an attributable, auditable act.
Task-mix shift: the routine tier (literature synthesis, boilerplate methods, standard stats, first-draft grants) is the part already automated; if that tier is stripped out, the residual job is hypothesis framing, failure diagnosis on contaminated or anomalous samples, and deciding which negative result kills a program. This raises the score only for the workers who keep the design tier — headcount effects are separate.
Biosecurity/dual-use rules that name a specific accountable human: e.g. the 2024 OSTP/USG Dual Use Research of Concern and Pathogens with Enhanced Pandemic Potential policy (effective May 2025) requires institutional review and named PI attestation; extension of nucleic-acid synthesis screening duties (EO 14110 framework) to require a named responsible scientist to certify sequence-design provenance, including AI-generated designs, would put personal sign-off on a licensed-equivalent human. Similarly IBC/IACUC protocols requiring a named investigator to attest that AI-drafted protocols were independently verified.
Growth in field and containment-dependent work that resists capital automation: BSL-3/4 handling, environmental sampling under permit (ESA Section 10 permits require named qualified biologists), and wildlife/agricultural surveillance where specimen condition judgment is in-person. Agency contracts specifying that permitted survey work be performed by a named qualified biologist on site raise this.
Journal and funder authorship rules that bar AI as author and require named human accountability for data (ICMJE, Nature policy) create attribution value but not buyer willingness to pay a premium; expert-witness and regulatory-testimony work is the only segment where a human is specifically purchased, and it is small.
The limit. No licensure path exists for the bulk of this category, so liability_shield gains are capped at protocol-attestation duties rather than true personal professional liability; and grant-funded headcount is set by budget, not by whether the residual judgment work is defensible.
| Washington-Arlington-Alexandria, DC-VA-MD-WV | 5,720 | $114,920 +16% |
| San Francisco-Oakland-Fremont, CA | 4,280 | $138,160 +40% |
| Boston-Cambridge-Newton, MA-NH | 2,630 | $110,410 +12% |
| Los Angeles-Long Beach-Anaheim, CA | 2,320 | $96,380 -3% |
| New York-Newark-Jersey City, NY-NJ | 1,970 | $111,380 +13% |
| San Diego-Chula Vista-Carlsbad, CA | 1,960 | $108,350 +10% |
| Seattle-Tacoma-Bellevue, WA | 1,880 | $108,580 +10% |
| Durham-Chapel Hill, NC | 1,390 | $98,700 +0% |
| San Francisco-Oakland-Fremont, CA | 4,280 | $138,160 +40% |
| San Jose-Sunnyvale-Santa Clara, CA | 610 | $133,790 +35% |
| Dayton-Kettering-Beavercreek, OH | 70 | $126,960 +28% |
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 43. 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.