Skip to content

Research report · Live data

Do Applicants Qualify for AI Training Roles?

Checked against the roles that suit them best, 27% of 180,543 checks found the people looking for AI training work a strong fit by 26 Oct 2026, and 91% of 7,248 members have at least one such role; when they fall short, the role most often asks for expertise in a profession or field.

Data window
29 Sep 2026 to 26 Oct 2026
28 days, UTC
Computed
27 Oct 2026, 10:22 UTC
Recomputed every day
Source
The profiles of the people looking for AI training work on AITraining.jobs (their LinkedIn profile, CV or own words, read by the site), the credentials they state or their profile shows, and the job alerts they subscribe to. Each fit check reads a member's profile beside a role's own posting.
Method
Version 1. Definitions
CC BY 4.0, how to cite

Every figure on this page as CSV and JSON, sent to your work email.

fit checks of 7,248 members against 4,131 roles, 29 Sep 2026 to 26 Oct 2026
180,543
of checks found a strong fit
27%
of members checked have at least one role they strongly fit
91%
of unlikely fits lack expertise in a profession or field
60%

Key findings

  1. 1

    Of 180,543 checks of members against the roles that suit them, 27% found a strong fit, 36% a possible one and 38% an unlikely one; 91% of the 7,248 members checked have at least one role they strongly fit.

    Figure 1
  2. 2

    When members asked about a role themselves, 38% of the checks found a strong fit, against 24% of the roles the site picked for them.

    Figure 2
  3. 3

    Members most often strongly fit Data Annotation & Labeling roles (55% of checks) and least often Research Studies & Surveys roles (3%).

    Figure 3
  4. 4

    84% of members with a licence or a doctorate have a role they strongly fit, against 91% of members with none of a licence, a doctorate or a master's.

    Figure 5
  5. 5

    When a member is an unlikely fit, the role most often asks for expertise in a profession or field (60%), years or seniority of experience (40%) and a language, or a level of it (35%) that their background does not show.

    Figure 7
  6. 6

    19% of checks name a language the member does not show; English is 26% of them, and Spanish (13%), Arabic (9%) and Japanese (6%) lead the rest.

    Figure 9
  7. 7

    The expertise members most often lack is in software and IT (29% of checks missing a field's expertise).

    Figure 10

Figure 1

How often members qualify

When a member is checked against a role that suits them, how often do they qualify?

VerdictChecksShare of checks
Strong fit
48,370
26.8%
Possible fit
64,130
35.5%
Unlikely fit
68,043
37.7%

Sample. 180,543 members' fit checks made 29 Sep 2026 to 26 Oct 2026, of 7,248 members against 4,131 roles.

A check is one member and one role, judged from the member's profile (their work, never their name or contact details) and the role's own posting: strong, possible or unlikely, with what fits and what is missing. Where a role is closed to the member's country, the verdict is unlikely, decided by the role's own rule.

Checks are made ahead for the roles the site lists for each member, and when a member opens a role and asks; so these are roles chosen for each member, not every role.

Members whose best it is: each member counts once, at the best verdict among their checks in the window.

No count below 3 people is shown, nor any share that would reveal one, so no figure can point at a person; a group too small to show on its own is counted in "Other".

Source. AITraining.jobs Research, Do Applicants Qualify for AI Training Roles?, live data.

Figure 2

When members ask, and when the site checks ahead

Do members who ask about a role themselves qualify more or less often than for the roles the site picks for them?

How the check was madeStrongPossibleUnlikely
The member asked about the role
37.9%
29.9%32.2%
Made ahead for the member's list
24.4%
36.7%38.9%

Sample. 180,543 members' fit checks made 29 Sep 2026 to 26 Oct 2026.

A member asks about a role from its page on the site; the site checks ahead the roles it lists for the member (their matches and their plan).

Source. AITraining.jobs Research, Do Applicants Qualify for AI Training Roles?, live data.

Figure 3

How often members qualify, by kind of work

For which kinds of AI training work do the people looking for it most and least often qualify?

Kind of workStrongUnlikely
Data Annotation & Labeling
54.7%
12.8%
Coding & Software Evaluation
32.6%
23.8%
Agentic Tasks & RL Environments
29.4%
28.7%
Multilingual & Localization
29.4%
44.7%
AI Evaluation & Rating (RLHF)
27.3%
23.0%
Transcription, Audio & Speech
26.2%
40.8%
Writing & Content
24.3%
30.8%
Red-Teaming & AI Safety
23.7%
23.8%
Robotics & Physical-AI Data
19.7%
25.7%
Law, Medicine & Finance
14.0%
49.0%
Math & STEM
11.4%
56.1%
Research Studies & Surveys
2.8%
36.3%

Sample. 180,543 members' fit checks made 29 Sep 2026 to 26 Oct 2026, by the kind of work of the role.

Each role is classified into one of the kinds of AI training work this site tracks, from its title and description.

Rows with fewer than 9 checks are marked as a small sample and not ranked: below that the average of the rows weighs more than a row's own checks (the weight fitted on the rows' own counts).

Source. AITraining.jobs Research, Do Applicants Qualify for AI Training Roles?, live data.

Figure 4

How often members qualify, by the field a role asks for

For roles in which professions and fields do applicants most and least often qualify?

Field the role namesStrongUnlikely
Software and IT
30.3%
25.8%
Writing and media
24.9%
34.3%
Languages and translation
23.3%
53.2%
Health and medicine
18.4%
52.7%
Finance and accounting
15.1%
47.3%
Data and AI work
14.9%
33.2%
Science, math and engineering
13.7%
50.0%
Law
13.2%
52.3%
Design, arts and audio
13.0%
53.8%
Business and management
12.1%
47.7%
Other
9.5%
61.8%
Teaching
7.0%
60.6%
The title names no profession
34.5%
33.9%

Sample. 180,543 members' fit checks made 29 Sep 2026 to 26 Oct 2026, by the field the role's title names.

A role is in a field when its title names one of the field's professions, by the rule the site's profession pages list roles with; a role naming two fields counts in both, and one naming none (most general rating, writing and language roles) is in the last row.

Rows with fewer than 24.6 checks are marked as a small sample and not ranked: below that the average of the rows weighs more than a row's own checks (the weight fitted on the rows' own counts).

Source. AITraining.jobs Research, Do Applicants Qualify for AI Training Roles?, live data.

Figure 5

How often members qualify, by the credential they hold

Do people with a licence, a doctorate or a master's qualify more often for AI training work?

Highest credentialWith a strong fitStrong share of checks
Licensed professional
84.5%
28.7%
PhD or doctorateSmall sample
85.5%
35.1%
Master's degree
87.5%
30.6%
None of these
90.7%
27.8%
Not known
92.6%
25.4%

Sample. The 7,248 members with a fit check made 29 Sep 2026 to 26 Oct 2026.

With a strong fit: the share of the group's members with at least one strong verdict among their checks in the window.

A credential is what the member stated on the site, else what they gave the candidate pool, else what their profile shows.

Rows with fewer than 212.5 members are marked as a small sample and not ranked: below that the average of the rows weighs more than a row's own members (the weight fitted on the rows' own counts).

No count below 3 people is shown, nor any share that would reveal one, so no figure can point at a person; a group too small to show on its own is counted in "Other".

Source. AITraining.jobs Research, Do Applicants Qualify for AI Training Roles?, live data.

Figure 6

How often members qualify, by their own field

Which fields' professionals most often find AI training work they strongly qualify for?

Member's fieldWith a strong fitStrong share of checks
Data and AI work
94.7%
30.7%
Languages and translation
92.8%
26.8%
Software and IT
92.2%
34.0%
Finance and accounting
91.7%
23.8%
Design, arts and audio
90.8%
26.0%
Law
89.4%
23.4%
Admin, support and service
88.8%
21.3%
Business and management
88.5%
19.7%
Teaching
88.2%
21.7%
Health and medicine
87.5%
22.4%
Writing and media
86.9%
22.2%
Science, math and engineering
86.1%
22.9%
Trades and operationsSmall sample
84.4%
17.9%
Not stated
82.4%
27.2%

Sample. The 7,248 members with a fit check made 29 Sep 2026 to 26 Oct 2026.

A member's field is the field of the best fit their profile reading names.

Rows with fewer than 143.4 members are marked as a small sample and not ranked: below that the average of the rows weighs more than a row's own members (the weight fitted on the rows' own counts).

No count below 3 people is shown, nor any share that would reveal one, so no figure can point at a person; a group too small to show on its own is counted in "Other".

Source. AITraining.jobs Research, Do Applicants Qualify for AI Training Roles?, live data.

Figure 7

What members most often lack

When a member falls short of an AI training role, what does the role ask for that their background does not show?

The role asks forShare of checksOf unlikely fitsOf possible fits
Expertise in a profession or field
44%
60%55%
Years or seniority of experience
27%
40%32%
Something else the role asks for
26%
33%35%
A language, or a level of it
20%
35%18%
A degree
16%
23%17%
A named tool or programming skill
11%
13%15%
Experience of AI training or annotation work
8%
7%13%
A professional licence or certification
5%
9%4%
Where they live, may work or when they can work
4%
7%4%

Sample. A simple random sample of 38,415 of the 180,543 members' fit checks made 29 Sep 2026 to 26 Oct 2026, with the gaps the judge wrote; a share of all checks is within 0.4 points of the whole window's, 19 times in 20.

Each check's gaps are the judge's own lines on what the role asks for and the member's background does not show. A check counts once for each kind its gaps name; a check with no gap counts for none, so the kinds do not add up to 100%.

A gap is read with the requirement report's own patterns (degrees, credentials, years, skills), the language list of the languages report, the professions of the profession pages, the programming languages and tools of the programming-languages report, and three phrasings only a gap uses: AI training or annotation work, a span or rank of experience, and where or when a member can work.

Patterns: AI training work \b(annotat\w*|label(?:l)?ing|label(?:l)?ers?|ai[- ]?(?:train\w*|tutor\w*|data|evaluat\w*|rater|model)s?|llms?|rlhf|content moderat\w*|moderation|raters?|prompt\w*|transcri\w*|evaluating (?:ai|models?|llms?))\b; experience (\b\d{1,2}\s*\+?\s*(?:[-–]\s*\d{1,2}\s*)?(?:years?|yrs?)\b|\b(?:one|two|three|four|five|six|seven|eight|nine|ten)(?:[- ]plus)?\s+(?:or more\s+)?years?\b|\b(?:senior|seniority|mid[- ]?senior|mid[- ]level|track record|proven|professional experience|years of)\b); where and when \b(not open in|open only|only open|based in|us[- ]based|resid\w*|located|location|time ?zones?|availability|available|full[- ]time|part[- ]time|hours? (?:a|per) week|on[- ]?site|in[- ]person|work authori[sz]ation|visa|citizen\w*|us[- ]specific|country|countries)\b.

Source. AITraining.jobs Research, Do Applicants Qualify for AI Training Roles?, live data.

Figure 8

What members lack, by kind of work

Which kinds of AI training work most often ask for a field's expertise, a language, a degree or experience the applicant does not have?

Kind of workField expertiseExperienceLanguageDegreeTool or skillAI training workLicencePlace or hours
Law, Medicine & Finance67%44%6%25%10%2%13%4%
Multilingual & Localization25%12%54%16%0%9%2%5%
Coding & Software Evaluation56%39%3%6%40%4%0%2%
Transcription, Audio & Speech21%6%45%3%0%19%1%3%
Data Annotation & Labeling17%12%8%4%5%19%0%2%
Math & STEM81%50%5%62%18%3%4%21%
AI Evaluation & Rating (RLHF)39%32%10%6%13%11%1%12%
Writing & Content36%22%16%13%4%2%1%5%
Red-Teaming & AI Safety55%3%4%2%9%10%2%2%
Agentic Tasks & RL Environments48%17%3%14%24%8%3%2%
Research Studies & Surveys47%20%0%21%0%2%16%1%
Robotics & Physical-AI Data16%6%7%1%5%28%1%3%

Sample. A simple random sample of 38,415 of the 180,543 members' fit checks made 29 Sep 2026 to 26 Oct 2026, with the gaps the judge wrote; a share of all checks is within 0.4 points of the whole window's, 19 times in 20.

Shares of each kind of work's own sampled checks.

Rows with fewer than 11.6 sampled checks are marked as a small sample and not ranked: below that the average of the rows weighs more than a row's own sampled checks (the weight fitted on the rows' own counts).

Source. AITraining.jobs Research, Do Applicants Qualify for AI Training Roles?, live data.

Figure 9

The languages members most often lack

Which languages do AI training roles ask for that the people applying do not show?

LanguageOf checks missing a languageOf all checks
English
25.9%
4.98%
Spanish
13.1%
2.53%
Arabic
9.3%
1.79%
Japanese
6.4%
1.24%
Portuguese
6.0%
1.15%
German
4.8%
0.92%
French
4.5%
0.86%
Hindi
3.6%
0.69%
Chinese (Mandarin)
3.5%
0.67%
Korean
3.5%
0.67%
Italian
2.6%
0.51%
Bengali
2.4%
0.45%
Dutch
2.1%
0.41%
Turkish
2.0%
0.38%
Russian
1.7%
0.33%
Tamil
1.4%
0.27%
Filipino (Tagalog)
1.4%
0.26%
Hebrew
1.4%
0.26%
Urdu
1.4%
0.26%
Vietnamese
1.3%
0.24%
Telugu
1.1%
0.21%
Indonesian
0.9%
0.18%
Swedish
0.8%
0.16%
Malay
0.7%
0.13%
Thai
0.6%
0.12%
Punjabi
0.6%
0.12%
Romanian
0.6%
0.12%
Persian (Farsi)
0.6%
0.11%
Danish
0.5%
0.10%
Norwegian
0.5%
0.10%
Greek
0.4%
0.08%
Polish
0.4%
0.08%
Cantonese
0.4%
0.08%
Marathi
0.4%
0.08%
AmharicSmall sample
0.3%
0.07%
GujaratiSmall sample
0.3%
0.06%
HungarianSmall sample
0.3%
0.06%
CzechSmall sample
0.2%
0.05%
MalayalamSmall sample
0.2%
0.05%
GeorgianSmall sample
0.2%
0.04%
CatalanSmall sample
0.2%
0.04%
YorubaSmall sample
0.2%
0.04%
SwahiliSmall sample
0.2%
0.04%
AfrikaansSmall sample
0.2%
0.03%
UkrainianSmall sample
0.2%
0.03%
KannadaSmall sample
0.2%
0.03%
AssameseSmall sample
0.1%
0.03%
GalicianSmall sample
0.1%
0.02%
FinnishSmall sample
0.1%
0.02%
Haitian CreoleSmall sample
0.1%
0.02%
OdiaSmall sample
0.1%
0.02%
BasqueSmall sample
0.1%
0.02%
MalteseSmall sample
0.1%
0.02%
NepaliSmall sample
0.1%
0.02%
BulgarianSmall sample
0.1%
0.01%
CroatianSmall sample
0.1%
0.01%
HausaSmall sample
0.1%
0.01%
AlbanianSmall sample
0.1%
0.01%
BurmeseSmall sample
0.1%
0.01%
KazakhSmall sample
0.1%
0.01%
LatvianSmall sample
0.1%
0.01%
LithuanianSmall sample
0.1%
0.01%
MāoriSmall sample
0.1%
0.01%
SerbianSmall sample
0.1%
0.01%
SlovakSmall sample
0.1%
0.01%
CebuanoSmall sample
0.0%
0.01%
JavaneseSmall sample
0.0%
0.01%
LuxembourgishSmall sample
0.0%
0.01%
MacedonianSmall sample
0.0%
0.01%
SomaliSmall sample
0.0%
0.01%
ArmenianSmall sample
0.0%
0.01%
SlovenianSmall sample
0.0%
0.01%
UzbekSmall sample
0.0%
0.01%
WelshSmall sample
0.0%
0.01%
AzerbaijaniSmall sample
0.0%
0.00%
EstonianSmall sample
0.0%
0.00%
KinyarwandaSmall sample
0.0%
0.00%
LaoSmall sample
0.0%
0.00%
NahuatlSmall sample
0.0%
0.00%
PashtoSmall sample
0.0%
0.00%
WolofSmall sample
0.0%
0.00%
ZuluSmall sample
0.0%
0.00%

Sample. A simple random sample of 38,415 of the 180,543 members' fit checks made 29 Sep 2026 to 26 Oct 2026, with the gaps the judge wrote; a share of all checks is within 0.4 points of the whole window's, 19 times in 20. 7,394 of the sampled checks name a language in a gap.

A gap names a language by the words the languages report reads in role titles. Where a role needs a language the member shows no sign of speaking, the judge's verdict is unlikely, whatever the rest of their background.

Rows with fewer than 27.3 sampled checks are marked as a small sample and not ranked: below that the average of the rows weighs more than a row's own sampled checks (the weight fitted on the rows' own counts).

Source. AITraining.jobs Research, Do Applicants Qualify for AI Training Roles?, live data.

Figure 10

The expertise members most often lack

In which fields do AI training roles ask for expertise the people applying do not show?

FieldOf checks missing a field's expertiseOf all checks
Software and IT
28.7%
12.6%
Science, math and engineering
20.2%
8.9%
Finance and accounting
13.6%
6.0%
Languages and translation
10.3%
4.6%
Health and medicine
9.1%
4.0%
Design, arts and audio
8.4%
3.7%
Business and management
7.9%
3.5%
Data and AI work
7.4%
3.3%
Teaching
5.7%
2.5%
Law
4.6%
2.0%
Other
3.3%
1.5%
Writing and media
2.6%
1.1%

Sample. A simple random sample of 38,415 of the 180,543 members' fit checks made 29 Sep 2026 to 26 Oct 2026, with the gaps the judge wrote; a share of all checks is within 0.4 points of the whole window's, 19 times in 20. 16,912 of the sampled checks name a field's expertise in a gap.

A gap names a field when it names one of the field's professions, by the rule the profession pages list roles with (a gap asking for 'legal domain expertise' names law).

Source. AITraining.jobs Research, Do Applicants Qualify for AI Training Roles?, live data.

For labs, data vendors and researchers

Buy the data behind this report

The figures above are a summary. Behind them are records we collect ourselves, every day:

  • 11,094 roles tracked, each with its history

    Every role on 56 job boards, read daily since 13 Jul 2026: its pay and place, and the first and last day it was open.

  • 37,580 apply clicks in 28 days

    Which roles people actually apply to, by field, pay and kind of work. It happens on this site, so nobody else can count it.

  • 39 lab and vendor hiring boards

    The job boards of 25 AI labs and 14 human-data vendors, read every day: who is hiring for human data, and who is hiring to buy it.

What you can buy

  • The records

    The rows behind any figure, with their history, not only the totals.

  • A regular feed

    The same figures every week or month, as files.

  • A custom cut or benchmark

    Your countries, languages or fields, or your own platform against the market.

Building AI and need the experts behind this data? (you contract them yourself) or (we deliver the expert work).

Investing in AI-data companies? Hiring radar is these records role by role, every day, as files.

Method and definitions

The profiles of the people looking for AI training work on AITraining.jobs (their LinkedIn profile, CV or own words, read by the site), the credentials they state or their profile shows, and the job alerts they subscribe to. Each fit check reads a member's profile beside a role's own posting. Every figure is computed by one database query over the window shown, the same query for the live report and for every edition. Each figure states its sample beneath it.

Member
A person with a profile on AITraining.jobs by the window's end: someone looking for AI training work who gave the site their LinkedIn profile, their CV or their own description of their work.
Fit check
One member against one role: the site reads the member's profile (their work, never their name or contact details) beside the role's own posting and answers strong, possible or unlikely, with what fits and what the role asks for that the background does not show. Checks are made ahead for the roles the site lists for each member and when a member asks about a role.
Gap
A line of the judge's on what the role asks for and the member does not show, read with the requirement report's own patterns, the languages report's words and the professions of the profession pages. A gap can be of several kinds.
Gap sample
What members lack is read from a simple random sample of the window's checks, 38,415 of them: the fewest for a share printed to a whole percent to be within half a point of the whole window's, 19 times in 20. A window with fewer checks is read whole.
Field
A member's field is the field of the best fit their profile reading names (software and IT, health and medicine, law, languages and translation, and the other groups the site's matching uses). A role is in a field when its title names one of the field's professions, by the rule the profession pages list roles with.
Highest credential
A professional licence (to practise as a physician, nurse, lawyer, pharmacist, psychologist, accountant, engineer and the like), else a research doctorate, else a master's degree, else none: what the member stated on the site, else what they gave the candidate pool, else what their profile shows. Not known when none of these says.
No cell points at a person
No count of fewer than 3 people is published, nor any share that would reveal one: with 3 or more, no one, not even a person counted, can tell who the others are. When one hidden cell could be worked out from a row's total, a second is hidden too.
Window
Live: checks made in the last 28 complete days, in UTC. Edition: one calendar month.

What this data cannot tell you

  • These are the people who use AITraining.jobs, not every person who does AI training work: the site's audience leans to the countries and fields it is found in.
  • A profile reading can miss what a profile does not say, and a credential nobody states is Not known, never guessed.
  • The roles checked are the ones that suit each member, so a share is of the work people would apply to, not of every role.
  • A verdict is a reading of what the profile and the posting say, not a platform's decision: a platform can accept someone the judge called a possible fit, or turn down a strong one.
  • A check made again for a newer judge replaces the earlier one, so a past window counts the verdicts as they now stand.

How the job feeds are read and roles are classified: site methodology. How fresh, complete and consistent the data is, checked every day: data quality and coverage.

Editions

The live report covers the last 28 complete days and changes every day. Each complete month is also published as an edition, frozen two days after the month ends, so a figure you cite from an edition never moves.

  • Live data (this page)Last 28 complete days, recomputed daily

Cite and reuse

Free to quote, chart and republish under CC BY 4.0 with a link to this page. Live figures change daily: cite a monthly edition for numbers that will read the same later, or keep the access date.

Citation

AITraining.jobs (2026). Do Applicants Qualify for AI Training Roles? (live data, 29 Sep 2026 to 26 Oct 2026). https://aitraining.jobs/research/do-applicants-qualify-for-ai-training-roles (accessed 27 Oct 2026)

BibTeX

@misc{aitrainingjobs_do_applicants_qualify_for_ai_training_roles_live,
  author = {{AITraining.jobs}},
  title = {Do Applicants Qualify for AI Training Roles? (live data, 29 Sep 2026 to 26 Oct 2026)},
  year = {2026},
  howpublished = {\url{https://aitraining.jobs/research/do-applicants-qualify-for-ai-training-roles}},
  note = {Accessed 27 Oct 2026},
}
Data: every figure as CSV (one line per value) and JSON, sent to your work email.Pay and open-roles data

More research

All 81 research reports