Neurodiversity 101: Neurofuturism, Neurodiversity and AI

What are the gains and potential challenges of AI in relation to neurodiversity?

I first wrote about this January 2023 in relation to education and then in my LinkedIn newsletter in August this year more about employment.

I recently spoke about this topic at the CIPD conference in Manchester as well in partnership with the great people at 10 to 3 who have developed with us a mini-training suite of materials we think will be very helpful for all workplaces and managers who are time-poor!

(Daniel Susskind was also talking about this at the conference from a broader sense regarding future employment -https://www.hrmagazine.co.uk/content/news/asking-if-ai-is-smarter-than-us-is-the-wrong-approach )

Here are my latest thoughts....

Where does AI and Neurodiversity come together?

Remember we have a lot of interconnected brain cells. We are messy and have different experiences and lives. We have different strengths and challenges and interests which all make solutions really complex when delivering services in bulk.

When it comes to addressing the needs of neurodivergent individuals, traditional approaches have often been overly simplistic and siloed because we don't have the capacity to make nuanced decisions or deliver personalisation especially at scale. We can see that present service delivery models are under pressure and not meeting current demands.

Could AI technologies bring a wave of potential benefits but what are the challenges at the same time?

“The person who follows the crowd will usually go no further than the crowd. The person who walks alone is likely to find himself in places no one has ever seen before.”― Albert Einstein

Potential benefits?

• Enhanced and equitable diagnostic approaches: AI offers the promise of more accurate and nuanced diagnostic tools, enabling healthcare professionals to better understand and support greater numbers of neurodivergent individuals. This could allow us to deal with the messiness that we grapple with. We need to remind ourselves that we already know that neurodevelopmental conditions are dimensional and people do not fall into the neat categorisations that are on offer. This could provide the means of delivering neurodevelopmental services that encompass all cognitive diversity rather than favouring some traits over others. Using Do-IT Profiler we are already able to capture other important factors such as life adversity. By using this data effectively we could target those in greatest need rather than targeting those who shout the loudest.We have been publishing this work showing the importance of gathering information that is not 'just' about neurodiversity in isolation.

• A paper published this week has also highlighted how looking at large data set providing is with new understanding about adversity and neurodiversity.
• Generative AI-based approaches can enhance digital accessibility, including speech-to-text and image-to-speech conversion, making information more accessible to a broader audience. This allows personal choice to happen. We need to move away in schools for example seeing that everyone needs to learn to write neatly. We have alternative means fo recording speech- to- text and adaptive typing and this needs to be seen as an option that does not have to wait to one leaves school (**3-5% of children have Developmental Coordination Disorder and handwriting is often a real challenge... why are we making some of these kids practice this skill when they could be focusing on other aspects of learning which will have a greater lasting impact?)
• Personalised interventions: AI can tailor interventions to the specific needs of each person. Whether it's adaptive learning in education or personalised therapies, AI has the potential to revolutionise neurodiversity support.
• Communication aids: For those who have different ways of communicating which may be different from neurotypical communication methods, AI-powered technology can serve as a bridge, making it easier for neurodivergent individuals to express themselves in the way that best works for them and for neurotypical individuals to check their communications have been fully understood.... good for all!
• Mental Health interventions: Many people who are neurodivergent have higher rates of mental health challenges.AI-driven non-medical interventions for conditions like anxiety and depression are becoming more accessible. We see this with an increase in use of 'wearables', AR and VR software.These tools can provide valuable support for individuals dealing with mental health challenges. Measuring changes in mood over time and in real time means we can respond to individuals making personalised and contextual suggestions.
• Flexible training: AI allows for adaptable training schedules, accommodating the individual's pace. This flexibility is especially important for some neurodivergent individuals who may require more time, more practice or unique approaches to learning and skill development. Being able to practice a skill in your own time, and at your own pace could aid and open u education and employment opportunities.

What are the challenges and risks in AI Implementation with regard to neuroinclusive practices?

While AI holds immense promise, there are several challenges and potential risks when it comes to ensuring neuroinclusivity:

• Neurodiversity: AI interactions need to be flexible to accommodate varied communication styles, emotional recognition, and speech patterns among those with a wide range of cognitive traits and characteristics.
• "Human biases are well-documented, from implicit association tests that demonstrate biases we may not even be aware of..... AI can help identify and reduce the impact of human biases, but it can also make the problem worse by baking in and deploying biases at scale in sensitive application areas." People with a wide range of lived experiences need to be involved in the design processes from the start otherwise we end up exaggerating the biases that already exist within the data sets being sampled.
• Facial differences and gestures: AI systems should not discriminate based on facial differences, asymmetry, or different gestures, ensuring facial recognition technology is inclusive. We have evidence of biases emerging in recruitment software algorithms e.g. CV scanning, automated video interviews (This AI is recognising voice patterns, word choices and may be checking facial expressions. There have been concerns around bias when a computer is analysing an applicant’s responses.),game based assessments and psychometric testing.
• Physical differences: AI systems must be designed to accommodate the unique variations in individuals with physical disability and create personalised solutions e.g. extending access to made- to- measure wheelchairs.
• Visual and hearing impairments: AI needs to provide alternatives to visual cues, making information and interactions accessible through auditory or tactile means ensuring they receive important commands and alerts effectively.
• Speech Impairment: Improved speech recognition technology and alternative methods of interaction are essential for those with speech impairments to effectively communicate with AI systems.
• Diverse communication styles : AI should embrace diverse communication styles and support various assistive devices to ensure equitable access for neurodiverse individuals.
• Digital literacy gaps: Ensuring that AI solutions are user-friendly and inclusive, especially for those with varying degrees of digital literacy, is crucial. Poverty can also have a double impact in terms of availability of the technology could create a widening bias.
• Bias and discrimination: AI systems must be carefully designed to avoid perpetuating biases, particularly in areas like employment, public services, and social recognition.If training data include biased human decisions or reflect historical or social inequities this can be exaggerated over time."Too often performance disparities are uncovered only after AI systems have been deployed and even after people have already experienced fairness-related harms" Researchers have developed technical ways of defining 'fairness', such as requiring that models have equal predictive value across groups or requiring that models have equal false positive and false negative rates across groups.
• Privacy concerns: Protecting the privacy of all individuals is paramount. AI systems must be developed and implemented in ways that respect personal data and ensures we gain informed consent. This means ensuring consent is understood and asked for in a way that people can engage with in order to make a decision.

While AI presents a host of opportunities for improving neuroinclusivity, addressing the challenges and mitigating potential risks is essential to create a more inclusive and equitable future for all.

***A BIG ASK***

We published the first Neurodiversity Index Report with City and Guilds Foundation in 2023 and now the survey is open again to get some really important benchmark data.

Please, please, please partake - we would love to see if the experience of neurodivergent employees in the workplace has changed or has training improved.?

Claim your City and Guilds digital credential

??You can take the survey here:https://lnkd.in/eDntbACt

Blog Author

I am Amanda Kirby, CEO of?Do-IT Solutions ?a tech-for-good company that delivers consultancy and guidance, training and web-based screening tools that have helped 10s of 1000s of people. We strive to deliver person-centered solutions relating to neurodiversity and wellbeing.

I am a mixed bag of experiences and skills and have 25+ years of working in the field of neurodiversity. I am a medical doctor, Professor, and have a Ph.D. in the field of neurodiversity; parent and grandparent to neurodivergent wonderful kids and am neurodivergent myself (bits of me I share!). I am current chair of the ADHD Foundation.

See the award-winning book?Neurodiversity at Work?Drive Innovation, Performance, and Productivity with a Neurodiverse Workforce and Neurodiversity and Education ?.

References:

https://www.weforum.org/agenda/2023/11/generative-ai-holds-potential-disabilities/

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10053494/

https://arxiv.org/pdf/2112.05675.pdf

https://www.edf-feph.org/understanding-artificial-intelligence-and-how-it-affects-the-disability-community/

https://europeanaifund.org/