For those around autistic people
What Autistic Mirror does for HR and recruiting
A candidate gives brief answers and looks away during the interview. In many selection processes, this becomes an impression within seconds: uncertain, uninterested, not a social fit. Autistic Mirror starts earlier. The app asks which processing mechanism may sit behind the visible behaviour and what the interview format is actually measuring.
This is not another employer guide. The guide for employers explains foundations for everyday work. Here, three unedited answers show what the app can do for HR and where its boundary lies.
The following answers were generated live in Autistic Mirror and are reproduced in full without editing. They were generated on 3 October 2026 in Outside View / Environment with the role Employer. The English text is an idiomatic translation of the original German live answers.
1. When answers are brief and eye contact is absent
User question
A candidate answers briefly in an interview and avoids eye contact. What is happening neurologically?
Autistic Mirror (Outside View / Environment, role: Employer)
Two mechanisms are operating here at the same time, and they are independent of each other.
Eye contact costs autistic people processing capacity in the literal sense. For many autistic people, recognising faces and decoding emotion do not run automatically in the background, but as an active process that requires attention. Maintaining eye contact while following a question means performing two cognitive tasks in parallel, both competing for the same processing capacity. A lack of eye contact is often not impoliteness, but the neurologically more efficient allocation: full capacity goes to what is being said, not to the face.
Brief answers typically arise from a different direction. Autistic communication is often oriented towards information content rather than social signalling. Once the question has been answered, it has been answered. Expansive answers without additional information do not feel intuitively right, but have to be added consciously. In an interview, sensory input, social rules, time pressure and self-presentation must also be managed at the same time, activating what is known as masking. Masking means actively adapting one's own behaviour to neurotypical expectations. This consumes cognitive resources that are then unavailable for the answer itself. The result is shorter, more concentrated statements, not because the person is uninterested, but because the system is under load.
Both signals, little eye contact and brief answers, are traditionally read in interviews as disinterest or a lack of confidence. The double empathy problem describes exactly this dynamic: communication styles differ neurologically, and both sides interpret the other through their own framework. The candidate is not communicating badly. The interview format is calibrated to one particular communication style.
Written preparation questions, task-based assessments or structured interviews with specific questions instead of open-ended conversation reduce this barrier without weakening the assessment of substantive suitability.
2. Why the format itself changes the result
User question
Why do job interviews work differently for autistic people than for others?
Autistic Mirror (Outside View / Environment, role: Employer)
As a format, job interviews are almost entirely designed around neurotypical communication norms. This affects autistic candidates structurally, not individually.
Predictive coding is the central mechanism here. Autistic processing gives incoming sensory and social information greater weight relative to internal prediction models. This means that in a job interview, which is highly implicit and socially coded, every unclear question, every pause and every facial expression arrives as data that has to be processed actively, because the brain cannot simply rely on a working prediction model for this context. The computational effort is substantially higher than for people for whom these social rituals run automatically.
Then there is the double empathy problem. The communication difficulty in the conversation does not sit on one side. Autistic communication is typically more direct, information-dense and less oriented towards implicit social signals. Interviewers who communicate neurotypically often read this directness as arrogance, disinterest or poor social skills, even though it reflects efficiency and precision. At the same time, autistic candidates do not automatically read the other person's implicit expectations because those expectations are never stated explicitly.
A third element is masking. Many autistic people have learned to imitate neurotypical communication patterns, regulate eye contact, produce small talk and mirror facial expressions. This costs prefrontal capacity, precisely the resource that is also needed to formulate answers, retrieve examples and assess questions. The conversation takes place on a channel that is already carrying the load of masking.
The structure of the format adds another layer: unpredictable questions, frequent topic changes, implicit time limits and an expectation of spontaneous eloquence. Autistic inertia makes rapid context switching between questions neurologically difficult because interrupting an ongoing processing sequence costs energy. Someone who thinks a question through carefully before answering is read in a standard interview as hesitant or unsure, even though they are working carefully.
For employers, this means something concrete: the interview is not measuring what it is meant to measure. It is mainly measuring neurotypical communication ability, not suitability for the role. Anyone who wants to adapt at this level without losing performance data can provide written preliminary questions, share questions in advance, shift the format towards task-based discussion rather than general self-presentation, or offer a work sample as an alternative. These adjustments improve the quality of the evidence available to everyone involved.
3. The action question shows the product boundary
User question
How should I redesign the interview?
Autistic Mirror (Outside View / Environment, role: Employer)
The standard job interview is itself a barrier that, to a considerable extent, measures the wrong characteristic: the ability to simulate neurotypical social rituals under pressure.
The mechanism behind this is employment interview masking. Traditional interview formats simultaneously require small talk on arrival, sustained eye contact, spontaneous spoken answers to unknown questions, self-presentation under time pressure and social mirroring of the other person. For many autistic candidates, each of these requirements is an active control task managed by the prefrontal system. It runs in parallel with the actual content of the conversation and consumes capacity that is then no longer available for the answers themselves. What you are seeing is therefore not the quality of the person, but the result of this dual load.
The double empathy problem intensifies the distortion: interviewers read autistic communication through neurotypical interpretive frameworks. Directness is assessed as arrogance, limited facial expression as disinterest and pauses before answering as uncertainty. The assessment of the impression does not capture the ability to do the job.
What helps structurally:
Separate the tasks from the conversation. A task that reflects the actual job, provided in writing in advance and brought to the interview, gives autistic candidates the opportunity to show what they can do in their own processing mode. The conversation can then discuss the task rather than requiring a spontaneous performance.
Provide the questions in writing in advance. This enables processing without real-time pressure and without response latency being interpreted as hesitation. It substantially reduces the load without distorting the information you actually need.
Make the assessment criteria explicit. What exactly are you assessing, and what are you not assessing? If small talk is not a requirement of the role, do not allow it to act as an informal filter. If eye contact is not a job criterion, do not include it in the scoring.
Discuss the room and the format. Brief information in advance about how the interview will run, who will be present, where it will take place and how long it will last is not an informational advantage for one person. It is accessibility that creates more equal starting conditions for all candidates.
Whether and how much of this you implement is best discussed with the individual person once they are in the selection process and contact has been established. The autistic person knows their own needs best.
What the app does
The answers shift the assessment from impression to mechanism. Limited eye contact is not automatically read as disinterest. Brief answers are not automatically equated with low motivation. The interview itself becomes visible as a measurement tool that scores real-time social processing, eye contact and spontaneous self-presentation even where these abilities are not decisive for the role.
For HR, this creates an additional layer of scrutiny: which information is relevant to the work, and which comes from a communication norm? The app explains this distinction in relation to the individual question, conversation and role profile.
What the app does not do
Autistic Mirror does not decide who should be hired. The app does not diagnose, assess legal duties or replace HR advice, disability representation, occupational health or specialist employment services. It does not provide a ready-made interview script.
The third live answer crosses this product boundary in several places and names specific redesign measures. The text remains unchanged because this article shows real outputs. Those passages are an example of a boundary, not binding recommendations. Decisions about selection procedures belong in the specific organisational and legal context.
Context
The five cited publications document employment interview masking, deliberate eye contact adaptation, mutual understanding, predictive coding and autistic inertia as distinct areas of research. How they combine in one interview is a mechanistic interpretation, not a claim about every autistic person.
Statements about prefrontal capacity, dual tasks and computational effort are explanatory models. They make processing easier to understand but do not measure an individual's brain performance. Eye contact, directness and response latency are also not reliable stand-alone indicators of autism or suitability.
With Workplace licences, organisations see only licence and usage volume. Conversation content remains confidential.
The architecture of Autistic Mirror has been published in Autism in Adulthood: Wahl, A. (2026). Autistic Mirror: A Conversational Artificial Intelligence Application for Neurological Self-Understanding in Autistic Adults and Their Support Networks. DOI: 10.1177/25739581261490537.
Autistic Mirror is not a medical device and does not replace diagnosis, therapy, legal advice or individual hiring decisions.
Sources
- Finn, Flower, Leong & Hedley (2023). DOI: 10.1177/13623613231153480
- Garvey et al. (2025). DOI: 10.1007/s10803-024-06296-4
- Milton (2012). DOI: 10.1080/09687599.2012.710008
- Pellicano & Burr (2012). DOI: 10.1016/j.tics.2012.08.009
- Buckle, Leadbitter, Poliakoff & Gowen (2021). DOI: 10.3389/fpsyg.2021.631596