Can technology help people with autism in the workplace?

In February 2021, the Office of National Statistics released figures showing that only 22 per cent of adults with autism in the UK are in any kind of employment. Only 16 per cent work full-time, according to the National Autistic Society (NAS).

These figures don’t come close to reflecting autistic people’s levels of capability in the workplace. Pioneers such as Einstein, Darwin, Newton, Mozart, Michelangelo are believed to have been autistic, albeit undiagnosed. Perhaps Steve Jobs too, some observers have suggested.

What the numbers do show, however, is a systemic failure to include and involve around 1 per cent of the UK population (according to NAS figures) in the workplace, cater for their needs, provide opportunities, and adequately use their talents.

NAS defines autism as a lifelong developmental disability that affects how people communicate and interact with the world. Society literature explains that people with autism have difficulties interpreting verbal and non-verbal language like gestures or tone of voice. Some have very good language skills, but struggle to understand sarcasm, figurative speech, idioms, or changes in intonation. Other challenges arise from a tendency to take things literally and difficulties understanding abstract concepts, or understanding others’ feelings and intentions.

“Autistic people have a strong need for predictability, it’s how they best process information, but the world of people is unpredictable,” says Professor Simon Baron-Cohen, head of Cambridge University’s Autism Research Unit. “You wouldn’t know how a conversation or social interaction might unfold, or how others will react to what you say.”

Professor Baron-Cohen - whose latest book on autism, ‘The Pattern Seekers’, came out last year - adds that people with autism can learn how to make social interaction more predictable. Technologies are emerging that may be able to assist them.

Researchers at UC San Diego’s Qualcomm Institute are currently designing an augmented-reality (AR) and virtual-reality (VR) coaching system that uses video games to help people with autism learn how to respond in various workplace scenarios and to become more aware of their own responses during conversation.

In one video game, participants control a spaceship by looking at it. If they don’t maintain eye contact with the ship on screen, the ship explodes. In another, participants choose between more diplomatic ways to turn down a request from a virtual work colleague to go out to lunch, when a straight ‘no’ or a blunt ‘don’t want to’ might be taken as rude. A third scenario sees participants pretend to wake up sick one morning and decide how to communicate that they will be unable to go to work, and who to inform.

“People can practise what to do, without consequences if they don’t get it right,” says Leanne Chukoskie, an associate research scientist and principal investigator on the project.

Pamela Cosman, professor of electrical and computer engineering at the UCSD Jacobs School of Engineering, explains that sensors in the AR headset measure participants’ non-verbal responses, such as head nodding in conversation to show engagement. “The participant will get feedback through a smart watch, via a decision algorithm.”

Chukoskie adds: “People are notified when they are standing too close, speaking too quickly for others to understand, or avoiding eye contact.”

In 2017, researchers at the Massachusetts General Hospital suggested that individuals with autism often avoid eye contact because it overstimulates a part of the brain responsible for emotional perception.

NAS explains that autistic people can also struggle to filter out less important information from what is important. Presented with too much information, a person with autism can easily become overloaded, to a point where they can’t process any further information.

Researchers from Vanderbilt University, Yale and Cornell Universities and the Georgia Institute of Technology, have been working on a job interview simulator to help people with their stress levels during the recruitment process – an obvious time when people are expected to maintain eye contact and take in lots of new information.

The Vanderbilt simulator tracks a person’s physical stress indicators – heart rate and sweat – during a mock interview. Observers can then identify the source of that stress –say, an interviewer asking too many open-​ended questions, or an interruption in proceedings.

A 2020 US study claimed that interruptions cause particular stress for people with autism because they have atypical activity in a part of the brain that regulates attention. To help with this, the Vanderbilt researchers have programmed a social robot to deliberately distract people during mock interviews and while engaged in other activities, such as watching TV. The researchers monitor any change in stress indicators and the robot records whether the person addresses the question and uses the appropriate social conventions.

“The short-term aim is that the autistic person better manages their stress during interview,” says Vanderbilt’s Nilanjan Sarkar, professor of mechanical engineering. “Long-term, we want to help employers and recruitment professionals find ways to gather the information they need in ways that create less anxiety and stress – maybe by breaking the question down into short steps, which are completed one at a time.”

Written communication can also be challenging for people with autism. When writing to work colleagues, many people use an informal shorthand full of buzz-phrases and colloquialisms, which draw on a shared set of values and assumptions to add meaning to what is being written.

“Because individuals with ASD [Autistic Spectrum Disorder, Professor Cosman’s preferred term] tend to take things very literally, they won’t necessarily get the hint if, say, they’re asked if they have an ‘update’ or if they could ‘take a look’ at a document,” Cosman says.
The UCSD researchers are designing an email tool to help avoid difficulties in written communication between individuals with ASD and their neurotypical co-workers.

The tool uses natural language processing technology to identify words or phrases that might not be clear or specific enough for individuals with autism. Filters scrutinise the email before it is sent and will highlight if detail or context is missing, or if the intent of the request is not clear, or if it is too open-ended. “The tool provides progressive clarification and offers substitutes,” Cosman explains.

Writing “could you work on this” implies that the speaker wants the listener to comply with the request now, but doesn’t say so explicitly, which can lead to confusion. The tool helps people spell out what they mean, for example: “Please stop what you’re doing and work on this. This is important, I’d like you to switch tasks now.”

Chukoskie says: “The individual with ASD then knows for sure what the expectation is and can ask for further guidance if needed.”

These assistive technologies are works in progress; their development has been delayed by the Covid pandemic. However, according to Dr Damian Milton, who lectures on intellectual and developmental disabilities at Kent University, using technology to ‘normalise’ autistic ways of being, around social interaction, is unwise. Dr Milton would like to see more focus on developing mutual understanding between autistic and neurotypical people.

“Social interaction involves people, unpredictability and mutual rapport-building,” he says. “We need something to help people understand and relate to each other and build this into the workplace system. Otherwise, we just imply that it’s the autistic person who has got it wrong.”

Professor Sarkar agrees. “We shouldn’t expect people with autism to change the way they communicate to something that makes them feel stressed and anxious,” he says.

Sarkar is of the view that people with ASD are a tremendous asset in the workplace, but he is frustrated that very few rise to the top of their professions. “Efforts shouldn’t end at getting autistic people a job; they should prosper too,” he says.

To this end, the Vanderbilt researchers are currently using cognitive modelling technologies to analyse autistic people’s visual reasoning abilities. The person wears an eye tracker containing sensors while doing a series of puzzles. Their results are then matched with a position they might excel in.

For a start, these may be roles that require concentration, memory, persistence, attention to detail and technical ability. People with autism often demonstrate above-average capabilities in these areas, NAS literature states.

A 2015 Cambridge University study found that engineers, physicists, inventors and mathematicians often exhibit autistic tendencies. Some, as already mentioned, to genius levels.

However, Professor Baron-Cohen, who led the Cambridge study, thinks that autistic people can just as easily excel as musicians, chefs or tailors – at anything, in fact, that requires them to work with rules, patterns, and systems.

Baron-Cohen explains that research shows a link between autistic genes and aptitudes for systemising. “Once a person with autism becomes familiar with a system, whether that is software, administrative or something else, they might quickly see how the system could be improved, or how a particular tool might work more effectively,” he says. “They might tweak one variable or parameter at a time, looking at what happens if they change it, whether the outcome is that the system works better.”

He adds: “Whereas some people might be frightened by new technologies, autistic people tend to jump in and start experimenting. This needs a lot of focus and patience, to run experiments over again, go to the ‘nth degree’ analysing the system.”

Baron-Cohen says that recognising patterns has been part of the process of human invention and innovation throughout history. “It was our ability to analyse and understand patterns that made humans scientific and technological masters of the planet 70,000 years ago,” he says.

If there’s a task or measure that can detail these unique abilities and approaches, then Professor Sarkar wants to find it. “We could then work out how strategies favoured by autistic people differ from those used by neurotypical people and introduce them into workplace and recruitment,” he says.

Chukoskie would just like to see more opportunities for people on the spectrum to learn the way they learn best, on the job. The apprenticeship model would work, she thinks, where newcomers observe the experts, over a period of time, while playing important support roles.

Unfortunately, to learn the game you have to be in the game. And February’s ONS figures show that 84 per cent of autistic adults certainly aren’t, when it comes to employment. “If you’re locked out of a particular world, there’s not much chance of figuring out how it works,” Chukoskie points out.