Why recruitment has become neuroinclusion’s slowest mover

Parul Singh is, by her own description, sitting in every seat at once. She is autistic and ADHD, and is a talent acquisition lead: meaning she is doing the hiring. She also founded neurodiversity consultancy Parallel Minds two years ago, which saw her advise organisations including NBCUniversal, Transport for the North and Siemens Healthineers. She recently spent a few months on the job market, seeking a permanent role following a contract’s expiration.

'Across all of that, I have seen how hiring processes are failing neurodivergent candidates from every side of the table,' she says.

Her argument is straightforward. Workplace neuroinclusion has made real progress in recent years across the employee lifecycle. Recruitment has not kept pace, and the gap is widening at precisely the wrong moment.

'We are in one of the most brutal job markets we have seen in a long time,' Singh says. 'Every candidate is fighting for their life. It is survival of the fittest, and hiring companies care less about inclusion when they are spoilt for choice.'

The consequence is that neurodivergent candidates are increasingly choosing not to disclose or request adjustments at all. The stakes have always been higher at interview than in post, but competition has sharpened the calculation.

The numbers behind it

The City & Guilds Neurodiversity Index 2025/2026 gives the picture some shape. Only 44% of employers provide interview adjustments to neurodivergent candidates. Nearly half, 49%, do not provide interview questions in advance even when a candidate asks. Just 31% have reviewed all their job descriptions for accessibility.

A separate YouGov and Evenbreak report found that 30% of disabled candidates believe employers hire disabled people only to hit a quota.

Singh is particularly critical of the Disability Confident scheme, launched to address the disability employment gap. 'All it has done is give employers a false sense of confidence,' she says. 'They join the scheme, display the badge and think the work is done.'

She points to UCL research by Kim Hoque finding the scheme had no measurable impact on employment outcomes, and to the government's own January 2026 announcement of an overhaul, in which its review conceded the scheme lacked teeth and risked being a symbolic gesture. 'Nearly a decade after it launched, that is a fairly damning verdict from the people who built it.'

When discrimination whispers

Singh distinguishes between discrimination that screams and discrimination that whispers. The loud version is rarer and easier to identify.

She cites a senior UX designer who posted publicly that a recruiter had worked out he was autistic from his advocacy work, then told him he would need to complete a capacity and IQ test to continue in the process. 'I am not exaggerating when I say my jaw dropped at how shockingly blatant it was.'

But it is the quiet version that concerns her more, 'because it hides behind the language of fairness and rigour.'

Her example is a man with 24 years at Microsoft and a career as a public speaker, rejected by a major employer and sent AI-generated feedback flagging 'communicating effectively' as a development area. He was told he came across as less confident than others, and advised to practise active listening and reflect on his communication style.

'Not only is this extremely patronising, it is wildly discriminatory to him as an autistic person,' Singh says. 'The system did not just filter him out. It explained the exclusion using deficit language, dressed up as innocent/helpful developmental feedback.'

Whether the discrimination screams or whispers, she argues, the underlying problem is identical: processes never designed with neurodivergent candidates in mind, which then treat the people they exclude as the problem.

Tech's progressive branding problem

Singh has spent most of her career in technology recruitment and she is candid about the sector’s self-image.

Some 53% of people in tech identify as neurodivergent, more than double the general population. Yet the assessment methods, she says, are among the least accommodating of any industry: live coding under observation, whiteboard exercises, rapid-fire technical questioning.

'Progressive branding and progressive process design are two very different things,' she says. 'Tech has generally nailed the first without doing the harder work of the second.'

She describes a software engineer who approached her for help. 

Recruiter conversations went well. First-round interviews went well. They fell apart, repeatedly, at the same stage: live coding. Their ADHD meant solving an unfamiliar problem while being observed caused her to freeze. They asked for an alternative assessment format. It was refused, on the grounds that the company needed to assess everyone consistently.

That word ‘consistency’ does considerable work, and Singh thinks it is worth challenging.

'Treating everyone the same is not the same as giving everyone an equal shot at showing what they can do,' she says. 'If one format reliably disadvantages a group of candidates, running everyone through it does not make it fair. It just makes the unfairness uniform. Consistency should live in what you are assessing, not necessarily how.'

Two things that need to be true at once

For Singh, a genuinely inclusive hiring process requires universal design and a clear adjustments pathway. Most organisations attempt one, usually by adding a line to job adverts encouraging candidates to ask for adjustments, and stop there.

She points to a 2023 study comparing the hiring experiences of autistic, non-autistic neurodivergent and neurotypical candidates. All three groups independently asked for the same things: less emphasis on social performance, more flexibility, more clarity about what to expect.

'Design for the group facing the most friction, and you tend to fix it for everyone else too.'

In practice, she builds a handful of shifts into every process she designs.

Supportive measures should be opt-out, rather than opt-in. For example, during interviews lasting over an hour, the interviewer should pause to break, but give the candidate an opportunity to continue if they wish.

Candidates should be asked what helps them perform at their best before they need it. A short preferences menu, asking whether they prefer direct or open questions, information in advance or in the room, more time or less, achieves considerably more than a vague invitation to request adjustments. 'Most candidates do not know what to ask for until they are already inside a process that is not working for them.'

Vague questions should be replaced with explicit ones. '“Tell me about yourself,” asks a candidate to guess what you actually want to know. “Talk me through two recent projects and your role in,” them does not.'

Logistical guesswork should be eliminated entirely: what type of interview it is, who the candidate will meet, and for on-site interviews, where to park, where the entrance is, how to find the office. 'All of this sounds trivial until you remember how much load uncertainty adds on top of an already high-pressure situation.'

And scoring should follow a structured matrix rather than interviewer instinct, so that communication and social differences are not weighted equally with capability to do the job.

'None of that requires knowing anything about a specific candidate's needs in advance,' Singh says. 'That is the point. It is designed in, not bolted on.'

Who carries the burden

The adjustments pathway matters because, as things stand, the work falls almost entirely on the candidate. One survey found that 78% of people had to initiate an adjustment request themselves, and only 10% found the process easy.

Singh's requirements for a functioning pathway are unremarkable, which is rather her point. Ask at every stage, not just the first. Never ask before a candidate can meaningfully answer: 'I have been asked what adjustments I need at application stage, before I knew what to expect from the interview process.' Explain the process properly so candidates can identify their own barriers. Help find the fix rather than simply logging the request. And treat it like any other logistical question, with no gratitude and no special treatment.

She describes a candidate who told her about the best interview process she had experienced. Before every stage she received a full brief on what to expect, removing the guesswork and much of the anxiety. Every brief explicitly invited adjustment requests, which meant that in her technical interview she felt able to ask the interviewer to read the task aloud, something she said she would never normally request. 

She disclosed her dyslexia and dyspraxia and it was simply acknowledged. She received detailed feedback after every stage. She got the job.

Her two previous processes had gone differently. One collapsed the moment she disclosed. The other sent task instructions during the interview itself.

'None of what made the good process work was expensive or complicated,' Singh says. 'It was proactive information, explicit permission to ask for what you need, and treating adjustment requests as completely unremarkable. That is genuinely most of it.'

The part that needs more scrutiny

Singh is clear that AI is receiving some scrutiny, but that there is not enough focus on its implications upon neurodivergent candidates and the extra layer of complexity it imposes.

The AI-generated rejection feedback she described earlier is a pattern. Screening tools increasingly assess communication style and cultural fit using language that measures neurological difference rather than job-relevant competence.

'It is a new, harder-to-see version of an old problem, because it comes wrapped in the appearance of objectivity,' she says. 'A human recruiter's bias can at least be questioned. A model's output gets treated as data.'

The question is being tested in the American courts. Derek Mobley, a Black disabled applicant in his 40s, alleges that Workday's AI screening tools filtered him out of more than a hundred applications on the basis of protected characteristics rather than job fit. A federal judge has allowed his disability discrimination claim to proceed. 

The case is among the first to test whether the company building the screening tool, rather than only the employer deploying it, can be held liable.

Closer to home, the Information Commissioner's Office engaged more than 30 UK employers on their use of automated decision-making in hiring. Many did not believe they were using automated decision-making at all, when in practice they were.

'That is not a small compliance footnote,' Singh says. 'If an organisation does not recognise it is making automated decisions, nobody is checking those decisions for bias either.'

What she would like to see is honesty. 'It is genuinely refreshing on the rare occasion a company tells candidates plainly how AI is used in their process, and what to expect from it. Transparency should be the baseline, not the exception.'

She extends that to herself. Despite using AI extensively as both professional and candidate, she draws a line: prep and idea generation, yes; writing her CV or interview answers, no.

'That distinction matters to me, and I think the recruitment industry, including me, needs to get more comfortable being specific about where that line sits, instead of leaving candidates to guess whether they are allowed to use it at all.'

Progress, and not enough of it

There is movement. Comparing the City & Guilds indexes across two years, 54% of organisations now say they have adapted their recruitment process, up from 51%. Job description reviews are up too, from 25% to 31%.

But Singh is clear that aggregate statistics and lived experience are not the same thing.

'I am in an unusual position to judge that, because right now I am sitting in every seat at once. I am neurodivergent myself, I am a consultant advising organisations on this, I am actively recruiting, and I am also, right now, a candidate.'

'That vantage point is exactly why I do not think the story is job done. It is: progress is real, and still not remotely enough.'

Parul Singh is a senior talent acquisition specialist in the tech sector and a Neurodiversity Advocate