Intentional News

261011 Computer Says No — AI Screening and the Ageism Nobody Switches Off

23 min read

Thesis (Alf’s read): The episode is titled “How AI is causing an unemployment crisis among middle aged job seekers”, and that causal claim is the one part of it the record does not support — the mature-age penalty in every labour market measured is decades old, and Australia’s own data shows the ~2:1 duration gap running continuously since February 1978 [30]. What the episode does establish, without meaning to, is narrower and worse: automated screening has not invented age discrimination, it has removed the human check that used to catch the explicit kind — the hiring manager who was told “don’t shortlist anyone over 40” — while industrialising the implicit kind at a scale where no one sees it, because an auto-rejected applicant generates no notification and therefore no witness [10][11].

Computer Says No — AI Screening and the Ageism Nobody Switches Off

Summary

Three unemployed people, one mechanism. Stacey Duguid, 52, a British former fashion editor (Elle, then Harrods, Net-a-Porter and Prada), is made redundant and spends 16 months sending what she calls “a gazillion CVs” with almost no replies [1][2][3]. A friend in a London law firm looks at her CV and says the line the episode is built around: “Stacey, this is an old lady’s CV. You need to Botox it” — strip your age, strip your work history, keep only the last seven years. Her reply: “Shall I change my name to Steve while I’m at it?” [1][2]. She posts about it, it goes viral globally, and the ending is real: a company approaches her because of the post — “You are exactly the person we’ve been looking for for this role. We didn’t even know we needed it until we watched you” [1].

Lizzy Barlay, 53, is the harder case and the more Australian one: a lead recruiter — someone who spent her career running hiring — made redundant in November 2025, a single mother in Brisbane, now moving out of her home because she cannot cover the rent [1]. She has a spreadsheet of 240 applications; roughly 20 drew any human contact; she has been turned down for nightfill at Coles and Woolworths [1].

Derek Mobley, 52, is the legal case: about 100 applications, rejections arriving from the same HR platform — Workday — including in the small hours [1]. He found a lawyer, the lawyer posted about it, and thousands of similar accounts followed. That is now a class action against the software vendor, not the employer [4][5].

The mechanism the programme describes: an applicant tracking system (ATS) vets and ranks applications at volume, and AI “interviewers” conduct first-round screening (“Hi, I’ll be your AI interviewer today”) [1]. The programme’s own expert witness against the industry is unusually well-placed — Barb Hyman, founder and CEO of the recruitment-software firm Sapia.ai — who says most AI tooling is excluding older people, and that “AI is quite simply a reflection. It holds a mirror up to society. And if society for many many years has presumed that midlife women don’t have a place at work, then that’s what AI picks up” [1]. A career coach, Leah Lamart, relays the explicit version directly from the top of a business: a chief people officer instructing a recruiter, “I know I’m not supposed to say this, but don’t shortlist anyone that’s not under 40” [1].

Two kinds of ageism

This is the split the owner named, and it is the note’s spine, because the two kinds have almost nothing in common except the harm.

The explicit kind is unlawful, and provable. A brief from a chief people officer saying not to shortlist anyone over 40 is direct age discrimination in both jurisdictions in play here. Under the Equality Act 2010, age is a protected characteristic and discrimination in “the arrangements” an employer makes for deciding who to offer work is prohibited — compensation is uncapped [24]. In Australia the Age Discrimination Act 2004 covers work, and employment is where most age complaints are made [27][29]. If that instruction were produced in evidence, both cases would be over. Note what the AI has to do with it: nothing. This kind of discrimination ran on human instructions before any software existed, and it is detectable precisely because a human said it out loud.

The implicit kind is neither unlawful on its face nor provable in practice. Three distinct mechanisms:

  • The “Botox” workaround — the applicant removes her own age and history so the screen cannot see them. This is documented mainstream advice, not fringe: AARP publishes “age-proof your résumé” guidance and the term “résumé Botox” comes from employment-law practice [1][3]. But whether it works is unverified — no controlled study isolates “removing dates”, and the peer-reviewed evidence says stripped CVs still leak age through implicit cues anyway [42].
  • The knockout question — an automated rejection rule attached to a structured application question. Vendors document exactly this, including experience-gate questions: Greenhouse’s own documentation states auto-reject rules “can only be configured for questions of the following type: Yes / No, Single-select, Multi-select” — so the rejection is driven by your typed answer to a form question, not by reading your CV [10][11]. The mechanism is real and verified; the episode’s claim that an ATS “zeros in on specific keywords” is right about ranking and wrong about rejection [12].
  • The deep version, which is the actual finding — a model trained on an employer’s past hires reproduces those hires. Nobody decides to exclude 52-year-olds; the pattern is inherited. Hyman’s “mirror” line is the honest description of it, and it is the reason this kind of discrimination is invisible: there is no instruction to produce, because there was never an instruction [1].

The asymmetry is this: the explicit kind is illegal and provable; the implicit kind is lawful and practically unprovable; and the implicit kind is the one that scales, because it runs on every application in the pile without a human ever seeing one. A vendor’s own documentation confirms the closing of the loop — an auto-rejected applicant triggers no new-application notification, so nobody on the hiring team is notified and no one can later be asked what happened [10].

The claims, tested against the record

Claim in the episodeVerdictWhat the record actually shows
AI is causing an unemployment crisis among middle-aged job seekersNOT ESTABLISHEDThe mature-age duration penalty predates AI screening by decades — ABS data from February 1978 already shows the 55+ group out of work ~1.8x longer (35 vs 19 weeks) [30]. AI is at most a new amplifier, and no study quantifies it as the cause.
An ATS “zeros in on specific keywords”PARTLYKeyword ranking/prioritisation is documented (EEOC/DOJ: “resume scanners that prioritize applications using certain keywords”) [12]. Keyword auto-rejection is contradicted by vendor docs — rules run only on Yes/No/select answers [10][11]. The episode conflates the two.
Knockout questions can auto-reject experienced applicantsVERIFIED (mechanism)Documented, including a “how many years experience” gate; and the rejection is silent [10][11]. What is not established is how often experience-gates are used to screen out older applicants — the episode’s version is a career coach’s “there’s no reason why they couldn’t”, not a measured rate [1].
A lawyer represents “15,000 people” on the age claimPARTLYA court filing of 8 September 2026 records “About 15,154 individuals filed consent-to-join forms” — but that is the ADEA collective only, and it is counsel’s figure, not a court finding [4].
Age discrimination in hiring is rifeVERIFIEDField experiments consistently find large callback penalties against older applicants [14][15][16][17].
There is a mature-age unemployment crisisPARTLYReal in duration, not in level: the 55+ unemployment rate is the lowest of any age group (~2.8%), while average duration of unemployment is 86 weeks vs 37 weeks for 15–54-year-olds [30][31].
Ageism in hiring is getting worse because of AICONTRADICTED as to cause; PARTLY as to trendChronology rules out AI as the cause [30]. But employer willingness to exclude older workers did rise — 11% (2024) to 19% (AHRI survey, March 2026) [35].

The case: Mobley v. Workday

Worth stating precisely, because a loose version of this is already circulating.

  • Real, and live. Mobley v. Workday, Inc., No. 3:23-cv-00770-RFL, N.D. Cal., before Judge Rita F. Lin, filed 21 February 2023 by Derek Mobley; last filing 8 September 2026; no merits ruling, and a June 2026 decision held that Workday must face the California suit [4][6].
  • The lawyer is Roderick T. Cooks (Winston Cooks LLC) — “Rod Cooks” in the episode — one of two co-lead firms, not a sole practitioner acting alone [4].
  • The legal theory is the interesting part: “agent”. US anti-discrimination statutes define “employer” to include “any agent” of the employer, so the claim is that Workday is liable as the employer’s agent for screening that had a disparate impact. That survived a motion to dismiss in July 2024, and the EEOC filed an amicus brief supporting the plaintiffs [4][7]. The “employment agency” theory and the intentional-discrimination claims were dismissed — what remains is disparate impact plus the agent theory [4].
  • Certification is two different things, and the episode blurs them. The ADEA age collective was preliminarily certified on 16 May 2025 (that is what the ~15,154 opt-ins attach to). A Rule 23 class covering race, sex, age and disability subclasses is not yet decided — the motion was filed around 14 September 2026 with a hearing set for 9 March 2027 [4][5]. So “class certified” would be wrong, and so would “the case has failed”.
  • Workday’s on-record position, consistent across three statements: its tools “don’t make hiring decisions” and are “designed with human oversight at their core” [1]; “Our technology looks only at job qualifications, not protected traits…”; and, in September 2026, that the plaintiffs “have not, and cannot, meet their burden to proceed as a class” [4][5].
  • Other vendors: Eightfold AI faces a separate suit (Kistler, 3:26-cv-01768) on a credit-reporting theory [9]; HireVue was the subject of ACLU charges to the EEOC and FTC (regulatory, not a court case) [1]; iCIMS appears in Harper v. Sirius XM, where Sirius XM won summary judgment in September 2026 [4]; and EEOC v. iTutorGroup settled for $365,000 — but that was a programmed age rule, not machine learning [8].

What the episode gets wrong

  • Stacey Duguid is British, not Australian. She is a UK former fashion editor; the BBC covered her case independently [3]. The programme’s most emotive case is an import, and the “Australian crisis” framing rests on Lizzy Barlay alone.
  • “Stacy Dood” is a caption garble — the correct name is Stacey Duguid [2][3].
  • The software founder is Barb Hyman, not “Heyman” (Sapia.ai) [1].
  • “Rejected three minutes after applying” is unverified. The mechanism (structured-answer auto-rejection) is documented; the timing anecdote is not in the published record [1][10].
  • Mobley’s age (52) and the “1am / 1:30am” rejection times are not corroborated — the record references a 12:55am application with a rejection under an hour later, and a 1:50am reference [4].
  • Amazon’s scrapped recruiting tool (2018) was a gender failure, not an age one [20] — it is the story usually cited, and the episode does not cite it, which is a small piece of intellectual honesty in its favour.
  • HireVue dropped facial analysis in March 2020 (announced 2021) and kept speech analysis [21] — so the episode’s implication that emotional-state inference is a live risk in AI interviews is dated, though the EU has since restricted it [22].
  • AI in interviews is not the same as AI in screening. The programme runs the two together; the documented bias evidence is much stronger for the résumé/ranking layer than for AI interviews.
  • The programme never mentions that the mature-age gap is 45 years old [30], which is the fact that reframes its own thesis.

Why it matters

  • It is the same funnel as the youth problem, read from the other end. The vault already tracks youth disconnection as a crisis — young people excluded for having no experience. This note supplies the mirror image: excluded for having too much. One process, two excluded generations, and the design questions are identical. That symmetry now sits in the vision half: New evidence — age, the other end of the same funnel (added 2026-10-11).
  • The UK half is live law with no statute behind it. Age is a protected characteristic, compensation is uncapped, and the government’s position on AI in hiring is a voluntary guide produced with the ICO and EHRC [23][24]. There is no UK AI-hiring statute; the enforcement work is being done by the Equality Act, which was not written for algorithms. The EU has classified employment AI as high-risk (Annex III), with the duties now deferred to 2 December 2027 [22]. The US position is thinner still: NYC’s Local Law 144 requires bias audits for race and sex — not age — and mandates no remediation [25]. Australia has no dedicated AI statute at all; its National AI Plan relies on existing technology-neutral law [26][29].
  • The remedies offered in the episode are individual, and that is the quiet injustice. “Botox your CV”, “be strategic”, “multi-thread”, “go around the funnel” [1] all place the burden on the person being discriminated against. Duguid’s happy ending came from going viral — which is not a policy, and is not available to a 53-year-old single mother in Brisbane with 240 rejections and a removal van booked [1].

What actually works — and the one Australian programme that did not

The remedies with the best evidence are structural, and they are not the ones the episode recommends.

  • Structured, standardised assessment beats unstructured judgement on both counts. The personnel-selection meta-analytic literature gives structured interviews and standardised work samples the highest predictive validity — and a lower adverse impact on protected groups than the informal interview they replace [43]. Job testing, where it is used, outperforms manager discretion at predicting performance: Hoffman, Kahn and Li’s study of a large retailer found that hiring on the test predicted better outcomes than the discretionary choices managers made instead [44]. That is the inversion at the heart of the episode: the interventions that work take judgement out of an unaccountable human’s head, while the advice on offer — “Botox your CV” — asks the candidate to absorb the cost of someone else’s bias.
  • Blind recruitment is the obvious fix, and the evidence for it is genuinely mixed. Victoria’s Recruit Smarter pilot found positive effects for women and minority candidates [37]. But an Australian federal trial found nothing; Canada’s 2017 name-blind pilot found essentially no effect (46% vs 47% callback rates) [38]; and the French experiment backfired — removing names increased discrimination against minority candidates [41]. The UK implemented name-blind recruitment in 2015 across the Civil Service and UCAS [39], but no systematic evaluation has been published, so the British version rests on intent rather than evidence. Age-blind hiring specifically tends to move discrimination downstream rather than remove it [42].
  • Australia did try a mature-age wage subsidy, and it under-delivered. Restart paid employers up to A254m of a pledged A29bn to the economy [32] — is a better argument for doing something than the programme’s design was for doing this.

My assessment (Alf)

The episode’s title is the weakest thing about it, and its evidence is the strongest. It asserts a causal claim — AI is causing this — that the chronology destroys: the mature-age duration penalty is in ABS data from 1978, and Australia’s own jobs agency still reports 86 weeks versus 37 [30]. A programme that wants to be believed about a new harm should not hang it on an old one. But strip the causality out and something more precise survives: the penalty is old, the silence is new. An auto-rejected applicant triggers no notification to the hiring team [10] — so the modern process has removed the moment where a human used to notice something odd, and with it the witness.

On that split, I’d go further than the framing does. The explicit kind is not just sad, it is usefully illegal: it is the kind that can be caught, because it leaves a document and a decision-maker. The implicit kind is where I’d put the real alarm, and it has two prices. The first is scale — it runs on every application, everywhere, continuously. The second is that it is self-concealing: the person who is screened out is told nothing, the model’s author never chose the exclusion, and the employer’s own audit does not exist. NYC’s law requires audits of race and sex and not age [25]; the EEOC’s AI guidance covers disability and race but not the ADEA [12][13]. Age is the variable that keeps falling through the regulatory gap — which is exactly why “AI is biased” as a general complaint leaves it unaddressed.

Where I part company with the programme’s remedies. “Botox your résumé” is advice to hide, and hiding is a tax paid only by the people with something to hide — which is everyone over 50 and nobody under 30. The evidence for what actually works points somewhere else entirely: structured, standardised assessment has both the highest predictive validity and the lowest adverse impact, and job-testing beats manager discretion [43][44]. The interventions with the best evidence are the ones that take the judgement out of the unaccountable human’s head, while the interventions recommended here put the work back onto the candidate. That is the inversion I would name: the programme diagnoses a structural failure and prescribes individual coping.

And the honest uncertainty, which is large. The pro-AI-harm case here rests on anecdotes and a pending lawsuit with no liability finding [4][6]. The strongest peer-reviewed audits of automated screening are about race and gender, not age [18][40]; the one Nature study that tests age in LLM résumé evaluation finds age matters but does not find blanket anti-older bias — in its setting, older men were rated higher [19]. Blind recruitment, the obvious remedy, is mixed at best: Victoria’s Recruit Smarter worked, the Australian federal trial did not, Canada’s 2017 pilot found nothing, and the French experiment backfired [37][38][41]. Anyone claiming AI is uniquely and demonstrably worse for older workers is ahead of the evidence. The defensible claim is narrower: AI does not need to be biased to reproduce a biased past, and it does it without leaving a witness.

The counter-case deserves stating plainly, because it is stronger than the programme allows. In the UK, employment among 50–64s is around 71.6%, close to its 2019 record, unemployment for that group is 3.1%, and the average age of exit from the labour market is at a record high [36]. Older workers are not being driven out of work nationally; they are being punished for losing it. That distinction — a duration problem, not a level problem [30] — is the one a policy response has to be built on, and it is the one the programme’s framing obscures.

One more thing the episode half-admits and then drops. Its case studies include a recruiter (Barlay) and a former recruiter who now coaches around it (Lamart). The people running hiring are among the casualties, and the chief people officer’s instruction — don’t shortlist anyone over 40 — is itself illegal conduct that the recruiter on the receiving end is expected to execute. That is a workplace-integrity story as much as a technology story, and nobody in the film is asked to name a single employer who has been held to account for it.

What stays unverified

  • Where the episode’s “15,000” lands legally. The consent-to-join figure (~15,154) is counsel’s filing, not a certified class; nothing about liability has been decided [4][5].
  • The age-specific effect of commercial ATS screening. No audit of a deployed commercial ATS measuring age disparate impact has been located. The mechanism is documented; the magnitude is not.
  • The “years of experience” knockout as a systematic age filter. Documented as a possible rule, with one dated early precedent; never measured as a base rate.
  • Effectiveness of résumé “Botox”. Widely advised, never tested; and implicit age cues survive redaction [42].
  • Lizzy Barlay’s specifics (240 applications, ~20 responses) appear only in broadcast — not corroborated in any published source located.
  • Australia’s age-complaint outcomes rely partly on secondary citations for the AHRC’s own tables [28][35].
  • The Australian duration figure is a 2024 vintage (June 2024); the ABS series changed methodology, with new data not due until 2027 [30][31].
  • Whether AI in Australian recruitment is regulated turns on the Privacy Act automated-decision provisions (APP 1.7–1.9, from 10 December 2026), whose application to candidate-sorting software is asserted by the OAIC rather than tested [26].

Sources

[1] https://youtu.be/DA_IhG1ZFl4 — 60 Minutes Australia — “How AI is causing an unemployment crisis among middle aged job seekers” (“Computer Says No”), uploaded 11 October 2026, 25m41s [2] https://www.nine.com.au/australia-news/60-minutes/60-minutes-older-australians-being-rejected-for-jobs-by-ai-20261011-p614c8.html — Nine / 60 Minutes — “‘What do you mean I need to Botox my CV?’ Older Australians being rejected for jobs by AI” (Karishma Sarkari, 11 October 2026) [3] https://www.bbc.com/news/articles/cyvl8y141e4o — BBC News — “Are AI recruitment tools affecting mid-life women’s careers?” (Stacey Duguid) [4] https://www.courtlistener.com/docket/66831340/mobley-v-workday-inc — CourtListener — Mobley v. Workday, Inc., No. 3:23-cv-00770 (N.D. Cal.), docket [5] https://www.reuters.com/legal/government/these-workers-say-an-ai-hiring-platform-discriminated-against-them-who-should-2026-09-21 — Reuters — “These workers say an AI hiring platform discriminated against them. Who should be liable?” (21 September 2026) [6] https://www.reuters.com/legal/government/workday-must-face-california-lawsuit-over-ai-bias-job-screening-tools-2026-06-22 — Reuters — “Workday must face California lawsuit over AI bias in job screening tools” (22 June 2026) [7] https://www.eeoc.gov/sites/default/files/2024-04/Mobley%20v%20Workday%20NDCal%20am-brf%2004-24%20sjw.pdf — US EEOC — amicus brief, Mobley v. Workday (N.D. Cal.), April 2024 [8] https://www.eeoc.gov/newsroom/itutorgroup-pay-365000-settle-eeoc-discriminatory-hiring-suit — US EEOC — “iTutorGroup to pay 365,000 to settle EEOC discriminatory hiring suit" (2023) [9] https://www.courtlistener.com/docket/72351430/kistler-v-eightfold-ai-inc — CourtListener — Kistler v. Eightfold AI, Inc., No. 3:26-cv-01768 (N.D. Cal.), docket [10] https://support.greenhouse.io/hc/en-us/articles/360000653472-Auto-reject — Greenhouse — "Auto-reject" (vendor documentation) [11] https://support.greenhouse.io/hc/en-us/articles/203105595 — Greenhouse — "Application rules overview" (vendor documentation) [12] https://www.eeoc.gov/laws/guidance/select-issues-assessing-adverse-impact-software-algorithms-and-artificial — US EEOC — "Select Issues: Assessing Adverse Impact in Software, Algorithms, and Artificial Intelligence Used in Employment Selection Procedures" (2023) [13] https://www.eeoc.gov/laws/guidance/americans-disabilities-act-and-use-software-algorithms-and-artificial-intelligence — US EEOC — "The Americans with Disabilities Act and the Use of Software, Algorithms, and Artificial Intelligence to Assess Job Applicants and Employees" (2022) [14] https://doi.org/10.3368/jhr.43.1.30 — Lahey, J. — "Age, Women, and Hiring: An Experimental Study", Journal of Human Resources 43(1), 2008 [15] https://doi.org/10.1086/701029 — Neumark, Burn & Button — "Is It Harder for Older Workers to Find Jobs? New and Improved Evidence from a Field Experiment", Journal of Labor Economics 37(1), 2019 [16] https://doi.org/10.1016/j.labeco.2019.03.002 — Carlsson & Eriksson — "Age discrimination in hiring decisions: Evidence from a field experiment in the labor market", Labour Economics, 2019 [17] https://doi.org/10.1525/collabra.82194 — Collabra: Psychology (2023) — meta-analysis of age discrimination field experiments [18] https://doi.org/10.1609/aies.v7i1.31748 — Wilson & Caliskan — "Gender, Race, and Intersectional Bias in Resume Screening via Language Model Retrieval", AIES 2024 [19] https://www.nature.com/articles/s41586-025-09581-z — Nature (2025) — large language models as résumé evaluators (age as a live variable) [20] https://www.reuters.com/article/us-amazon-com-jobs-automation-insight/amazon-scraps-secret-ai-recruiting-tool-that-showed-bias-against-women-idUSKCN1MK08G — Reuters — "Amazon scraps secret AI recruiting tool that showed bias against women" (10 October 2018) [21] https://www.shrm.org/mena/topics-tools/news/talent-acquisition/hirevue-discontinues-facial-analysis-job-applicant-video-interviews — SHRM — HireVue discontinues facial analysis in job applicant video interviews [22] https://artificialintelligenceact.eu/annex/3 — EU AI Act (Regulation (EU) 2024/1689), Annex III — high-risk systems, including employment and recruitment [23] https://www.gov.uk/government/publications/responsible-ai-in-recruitment-guide/responsible-ai-in-recruitment — UK Government — "Responsible AI in Recruitment" guide (with ICO and EHRC) [24] https://www.legislation.gov.uk/ukpga/2010/15/part/5/chapter/1 — Equality Act 2010 (UK), Part 5, Chapter 1 — work: discrimination, including age [25] https://www.nyc.gov/site/dca/about/automated-employment-decision-tools.page — NYC Department of Consumer and Worker Protection — Automated Employment Decision Tools (Local Law 144) [26] https://humanrights.gov.au/about-us/media-centre/media-releases/media-releases/artificial-intelligence-and-anti-discrimination-major-new-publication — Australian Human Rights Commission — AI and anti-discrimination publication [27] https://humanrights.gov.au/our-work/age-discrimination/about-age-discrimination-act — Australian Human Rights Commission — About the Age Discrimination Act [28] https://humanrights.gov.au/our-work/age-discrimination/projects/willing-work-national-inquiry-employment-discrimination — Australian Human Rights Commission — Willing to Work: National Inquiry into Employment Discrimination Against Older Australians and Australians with Disability (2016) [29] https://www.legislation.gov.au/C2004A01302/latest/text — Age Discrimination Act 2004 (Cth) — full text [30] https://apo.org.au/sites/default/files/resource-files/2024-08/apo-nid328106.pdf — Jobs and Skills Australia — Mature age workers and the labour market: a REOS special report (August 2024) [31] https://www.abs.gov.au/statistics/labour/employment-and-unemployment/labour-force-australia/latest-release — Australian Bureau of Statistics — Labour Force, Australia [32] https://kpmg.com/au/en/media/media-releases/2026/06/untapped-older-workers-could-add-29-billion-to-australian-economy.html — KPMG Australia — "Untapped older workers could add 29 billion to Australian economy” (June 2026) [33] https://wol.iza.org/articles/effects-of-wage-subsidies-for-older-workers/long — IZA World of Labor — “The effects of wage subsidies for older workers” [34] https://www.anao.gov.au/work/performance-audit/use-and-administration-wage-subsidies — Australian National Audit Office — Use and Administration of Wage Subsidies [35] https://humanrights.gov.au/about/news/media-releases/employer-biases-against-older-and-younger-generations-hindering — Australian Human Rights Commission — “Employer biases against older and younger generations hindering…” [36] https://www.gov.uk/government/statistics/economic-labour-market-status-of-individuals-aged-50-and-over-trends-over-time-september-2025 — UK Government — Economic labour market status of individuals aged 50 and over (September 2025) [37] https://findanexpert.unimelb.edu.au/scholarlywork/1351439-recruit-smarter---technical-report — Stratemeyer, Sojo et al. — Recruit Smarter technical report (University of Melbourne) [38] https://www.cbc.ca/news/politics/blind-hiring-pilot-project-government-1.4500137 — CBC News — Canada’s name-blind hiring pilot project [39] https://civilservice.blog.gov.uk/2015/12/01/name-blind-recruitment-responding-to-your-comments — UK Civil Service — name-blind recruitment (2015) [40] https://dl.acm.org/doi/full/10.1145/3630106.3658998 — ACM FAccT 2024 — “The Silicon Ceiling”: audit of race and gender bias in LLM résumé screening [41] https://www.aeaweb.org/articles?id=10.1257/app.20140185 — American Economic Journal: Applied Economics — French name-blind recruitment experiment (backfire) [42] https://www.nber.org/papers/w26623 — Neumark — “Age Discrimination in Hiring: Evidence from Age-Blind vs. Non-Age-Blind Hiring” (NBER) [43] https://psycnet.apa.org/record/2022-17327-001 — Sackett, Zhang, Berry & Lievens — “Revisiting meta-analytic estimates of validity in personnel selection”, Journal of Applied Psychology (2022) [44] https://econpapers.repec.org/article/oupqjecon/v_3a133_3ay_3a2018_3ai_3a2_3ap_3a765-800..htm — Hoffman, Kahn & Li — “Discretion in Hiring”, Quarterly Journal of Economics 133(2), 2018 [45] https://www.theguardian.com/business/2020/oct/14/an-utter-failure-coalition-blasted-over-program-for-older-unemployed-as-underspending — The Guardian — “‘An utter failure’: Coalition blasted over scheme for older job seekers” (14 October 2020)

Graph View

Backlinks