Key findings

1
Multilingual & Localization drew the most apply clicks: 436 of 2,351 (18.5%), followed by AI Evaluation & Rating (RLHF) (17.9%) and Transcription, Audio & Speech (14.8%).
Figure 1
2
Per open role, Agentic Tasks & RL Environments drew the most applicant interest: 0.71 apply clicks a week, 3.3x the average role. Law, Medicine & Finance drew the least: 0.09 a week (0.41x).
Figure 2
3
AI Evaluation & Rating (RLHF) took 17.9% of apply clicks with 9.7% of open-role time. Law, Medicine & Finance held 23.8% of open-role time but drew 9.7% of clicks.
Figure 2
4
A role in its first 7 days drew 1.07 apply clicks a week; a role listed for more than 90 days drew 0.09 (the new role drew 11.7 times as many).
Figure 4
5
Roles advertising $100/hr or more drew 0.25 apply clicks per open role a week; roles under $20/hr drew 0.22 (the higher-paid role drew 13% more).
Figure 3
6
The most-clicked 10% of open roles took 79.5% of apply clicks; 78.3% of the 4,299 roles open during the window drew none.
Figure 6

Figure 1

Apply clicks by kind of work

Which kinds of AI training work do people apply to?

Kind of work	Apply clicks	Share of clicks	Roles applied to
Multilingual & Localization	436	18.5%	218
AI Evaluation & Rating (RLHF)	420	17.9%	138
Transcription, Audio & Speech	347	14.8%	132
Data Annotation & Labeling	247	10.5%	61
Law, Medicine & Finance	228	9.7%	121
Coding & Software Evaluation	207	8.8%	74
Math & STEM	164	7.0%	66
Agentic Tasks & RL Environments	122	5.2%	41
Writing & Content	67	2.8%	31
Robotics & Physical-AI Data	50	2.1%	21
Red-Teaming & AI Safety	40	1.7%	17
Research Studies & Surveys	23	1.0%	13

Sample. 2,351 apply clicks on a specific role, from 1 Sep 2026 to 30 Sep 2026.

A further 1,787 clicks went to a platform's own sign-up page rather than to one role, so they have no kind of work and are not in this figure.

Roles applied to: distinct roles that received at least one of these clicks.

Source. AITraining.jobs Research, Applicant Demand for AI Training Work, September 2026 edition.

Figure 2

Applicant demand per open role, by kind of work

Which kinds of work are oversubscribed, and which are short of applicants?

Kind of work	Clicks per role a week	Index	Clicks	Avg open roles	Click share	Open-time share
Agentic Tasks & RL Environments	0.71	3.30x	122	40	5.2%	1.6%
Data Annotation & Labeling	0.54	2.50x	247	107	10.5%	4.2%
Writing & Content	0.44	2.05x	67	35	2.8%	1.4%
Research Studies & SurveysSmall sample	0.42	1.96x	23	13	1.0%	0.5%
AI Evaluation & Rating (RLHF)	0.40	1.85x	420	246	17.9%	9.7%
Robotics & Physical-AI Data	0.36	1.66x	50	33	2.1%	1.3%
Transcription, Audio & Speech	0.28	1.28x	347	293	14.8%	11.5%
Coding & Software Evaluation	0.22	1.01x	207	223	8.8%	8.7%
Math & STEM	0.19	0.89x	164	200	7.0%	7.9%
Red-Teaming & AI Safety	0.16	0.74x	40	58	1.7%	2.3%
Multilingual & Localization	0.15	0.68x	436	691	18.5%	27.2%
Law, Medicine & Finance	0.09	0.41x	228	607	9.7%	23.8%

Sample. 2,351 apply clicks on 4,299 roles open at some point from 1 Sep 2026 to 30 Sep 2026.

Clicks per role a week = apply clicks / (days the kind's roles were open / 7). It measures competition for each opening, not the size of the market: a kind with few roles and steady interest reads high.

Index: the kind's rate divided by the rate across all open roles. 2.00x means twice the applicant interest of the average role. Click share and open-time share are the kind's share of all role clicks and of all open role-days.

Rows marked as a small sample have fewer than 30 clicks or fewer than 3 roles open on an average day.

Source. AITraining.jobs Research, Applicant Demand for AI Training Work, September 2026 edition.

Figure 3

Applicant demand by advertised hourly pay

Do higher-paying roles attract more applicants?

Advertised pay	Clicks per role a week	Index	Clicks	Avg open roles
Under $20/hr	0.22	1.01x	193	207
$20 to $40/hr	0.20	0.95x	431	491
$40 to $60/hr	0.33	1.55x	314	219
$60 to $100/hr	0.20	0.94x	618	711
$100/hr or more	0.25	1.15x	336	318
No hourly rate stated	0.18	0.83x	459	600

Sample. 2,351 apply clicks on a specific role, from 1 Sep 2026 to 30 Sep 2026.

Pay is the hourly figure the posting advertises (the middle of a stated range). Roles paid per task or per project with no hourly figure are under 'No hourly rate stated'.

Source. AITraining.jobs Research, Applicant Demand for AI Training Work, September 2026 edition.

Figure 4

Applicant demand by how long a role has been listed

How quickly does interest in a new role fade?

Listed for	Clicks per role a week	Index	Clicks	Avg open roles
First 7 days	1.07	4.96x	942	206
7 to 30 days	0.24	1.09x	611	604
30 to 90 days	0.12	0.57x	406	778
More than 90 days	0.09	0.42x	366	934
Listing date unknownSmall sample	0.26	1.22x	26	23

Sample. 2,351 apply clicks on a specific role, from 1 Sep 2026 to 30 Sep 2026. A role's open days are split across the bands it passed through.

Age runs from the listing date the platform's job board states, or else from the day the role first appeared in the feed. Roles that came in with a platform's back catalogue and state no date are under 'Listing date unknown'.

AITraining.jobs ranks recently listed roles higher in its default order, which adds to the first-week figure.

Source. AITraining.jobs Research, Applicant Demand for AI Training Work, September 2026 edition.

Figure 5

Applicant demand by who can apply

Do roles open worldwide draw more applicants than roles limited to some countries?

Location rule	Clicks per role a week	Index	Clicks	Avg open roles
Open worldwide	0.29	1.35x	1,307	1,047
Worldwide except listed countries	0.26	1.20x	212	191
Only listed countries	0.15	0.69x	831	1,307
No rule statedSmall sample	0.28	1.29x	1	1

Sample. 2,351 apply clicks on a specific role, from 1 Sep 2026 to 30 Sep 2026.

AITraining.jobs shows each visitor the roles open in their country first and hides roles their country cannot apply to unless they choose another country, so roles limited to listed countries are seen by fewer visitors.

Source. AITraining.jobs Research, Applicant Demand for AI Training Work, September 2026 edition.

Figure 6

How concentrated applicant attention is

Do apply clicks spread across many roles or pile onto a few?

Roles	Number of roles	Share of apply clicks
Most-clicked 1% of roles	43	27.2%
Most-clicked 5%	215	60.6%
Most-clicked 10%	430	79.5%
Most-clicked 20%	860	98.5%

Sample. 2,265 apply clicks on 4,299 roles open at some point from 1 Sep 2026 to 30 Sep 2026, the roles ranked by the clicks each received (roles with none included).

21.7% of these roles (933) received at least one apply click.

Source. AITraining.jobs Research, Applicant Demand for AI Training Work, September 2026 edition.

Figure 7

Apply clicks per week

Is applicant activity rising or falling?

Week	Apply clicks	Per day
1 to 6 SepPartial week	494	82
7 to 13 Sep	826	118
14 to 20 Sep	996	142
21 to 27 Sep	1,275	182
28 to 30 SepPartial week	547	182

Sample. All human apply clicks on AITraining.jobs, 1 Sep 2026 to 30 Sep 2026, by ISO week (Monday to Sunday, UTC).

Weekly totals grow with AITraining.jobs's own audience as well as with the market. To compare kinds of work across weeks, read the shares in the next figure.

A week the month cuts is labelled with the days it covers and marked as partial.

Source. AITraining.jobs Research, Applicant Demand for AI Training Work, September 2026 edition.

Figure 8

Share of apply clicks by kind of work, week by week

Is interest shifting between kinds of work?

Week	Multilingual	AI Evaluation	Transcription & Audio	Data Annotation	Domain Experts	Coding	Other
1 to 6 Sep	24%	14%	14%	6%	12%	9%	22%
7 to 13 Sep	21%	19%	14%	7%	10%	8%	21%
14 to 20 Sep	19%	16%	11%	14%	10%	11%	20%
21 to 27 Sep	18%	15%	18%	12%	8%	8%	21%
28 to 30 Sep	12%	27%	14%	11%	11%	10%	14%

Sample. Apply clicks on a specific role, 1 Sep 2026 to 30 Sep 2026, by ISO week. Each row adds up to 100% (rounded).

Columns are the six kinds of work with the most clicks over the whole period.

Source. AITraining.jobs Research, Applicant Demand for AI Training Work, September 2026 edition.

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:

9,572 roles tracked, each with its history
Every role on 29 job boards, read daily since 13 Jul 2026: its pay and place, and the first and last day it was open.
6,200 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.

Method and definitions

Apply clicks made on AITraining.jobs, joined to the role clicked and the days each role was open. 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.

Apply click: A click on an apply button on AITraining.jobs that opens a platform's application page. Made by a person: automated traffic is set aside as it is recorded. Counted per click, not per person.
Role-level click: An apply click on one specific role. Clicks on a platform's general sign-up link are platform-level and have no kind of work, pay or listing age.
Open role-days: For every role, the days of the window it was listed, summed. A role open the whole of a 28-day window adds 28.
Clicks per open role a week: Role-level apply clicks divided by open role-days, times 7.
Demand index: A group's clicks per open role a week divided by the same rate across all open roles. Above 1, the group's roles draw more applicant interest than the average role.
Kind of work: Each role is classified from its title and description into one of the twelve kinds of AI training work this site tracks, from AI evaluation and coding to multilingual and audio work. Every kind links to its own page.
Listing date: The date the platform's job board gives for the posting, or else the first day the role appeared in the feed. A role that arrived with a platform's back catalogue and gives no date has no known listing date.
Location rule: Read from each posting's location text and, where that names no country, from what the posting's own description states: open worldwide, worldwide except listed countries, only listed countries, or no rule stated.
Window: Live: the last 28 complete days, in UTC. Edition: one calendar month, in UTC.

What this data cannot tell you

Clicks measure interest among the people who use AITraining.jobs, not across the whole market.
What is shown first gets clicked more. The default order ranks recently listed roles, roles that state their pay and roles at partner networks higher, and each visitor sees the roles open in their country first.
An apply click starts an application. It is not a submitted application, an acceptance or a hire.
One person can click several roles, so the figures count clicks, not people.

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 dataLast 28 complete days, recomputed daily
September 2026 editionFrozen 10 Oct 2026
August 2026 editionFrozen 10 Oct 2026

Cite and reuse

Free to quote, chart and republish under CC BY 4.0 with a link to this page. This edition is frozen, so its figures will read the same later.

Citation

AITraining.jobs (2026). Applicant Demand for AI Training Work, September 2026 edition. https://www.aitraining.team/research/applicant-demand/2026-09

BibTeX

@misc{aitrainingjobs_applicant_demand_2026_09,
  author = {{AITraining.jobs}},
  title = {Applicant Demand for AI Training Work, September 2026 edition},
  year = {2026},
  howpublished = {\url{https://www.aitraining.team/research/applicant-demand/2026-09}},
}

Data: every figure as CSV (one line per value) and JSON, sent to your work email.Pay and open-roles data

More research

All 78 research reports