AI through the lens of neurodiversity
Notes: This article distinguishes the term ‘neurodiverse’ from ‘neurodivergent,’ which refers to an individual whose brain type is atypical. By definition, ‘neurodiversity’ includes neurotypical individuals as well; therefore, it is important to make this distinction.
Every now and then, we encounter events that help us extend our perceptions of the world. They can come from interactions with other humans, experiences that life thrusts upon us, such as an unfortunate illness, or even reading a book. One such book that has been particularly eye-opening and intellectually fulfilling for me over the years was My Stroke of Insight by Dr. Jill Bolte Taylor.
The author experienced a rare form of stroke in 1996. In this book, she provides a unique account of her personal journey of recovery and the neuroscience behind her experience. After waking up with a severe headache, Dr. Taylor realized she was in the midst of a stroke when her bodily functions and cognitive abilities began to deteriorate. Surprisingly, as the left hemisphere of her brain went offline, she also experienced a heightened state of peace and connectivity with the world, linked to the right hemisphere.
What this book did for me was to help realize that the way I perceive the world is just one of countless possibilities. Our brains are truly marvelous — One of the quotes from the book illuminates this:
“I had lost touch with much of the physical three-dimensional reality that surrounded me. My body was propped up against the shower wall and I found it odd that I was aware that I could no longer clearly discern the physical boundaries of where I began and where I ended. I sensed the composition of my being as that of a fluid rather than that of a solid. I no longer perceived myself as a whole object separate from everything. Instead, I now blended in with the space and flow around me.”
Even keeping in mind that her experience was the most dramatic and not something most people will ever go through in their lifetimes, this ‘stroke of insight’ is deeply profound. The story underscores how even minor alterations in the brain can significantly shift our worldview, and it’s not hard to extend this finding to imagine that every single person you see on the street could be perceiving the world in their unique ways too.
Consider, for instance, a deaf individual; the absence of hearing and the use of sign language likely heighten their spatial awareness (as documented in Oliver Sacks’ Seeing Voices). This person’s interactions with the world must be different from hearing individuals, with a more developed part of the brain responsible for spatial awareness. It’s a captivating thought — we can imagine billions of unique eyes viewing the world, even if the differences among most can be said to be minor. When these differences become clinically significant, this is where neurodivergence enters.
Neurodiversity and AI
Research indicates that about 15–20 percent of the population is neurodivergent. This includes up to 10 percent of people diagnosed with dyslexia, 6 percent with dyspraxia, 5 percent with ADHD, and 1–2 percent with autism. (And of course, there are a number of other forms of neurodivergence too.) Neurodivergence offers a multitude of perspectives on perception but also makes us question the traditional notions of intelligence. Historically, intelligence has been narrowly defined and measured through tools like IQ tests. However, once we embrace the concept of neurodivergence, intelligence is no longer a linear scale but becomes a multi-dimension of capabilities. A person with dyslexia, for instance, might face challenges with reading and writing but excel in spatial reasoning. Similarly, an individual with autism might possess extraordinary memory or pattern recognition skills. This diversity in neural wiring highlights the multifaceted nature of intelligence, which cannot be confined to a single metric.
This nuanced understanding of intelligence can be crucial to our understanding of artificial intelligence (AI). We tend to project our notion of intelligence onto AI, often constraining it within a singular, human-centric paradigm. As a result, media debates center around what AI can and cannot do, in comparison with human capabilities. Does AI have humor? Can it genuinely empathize? Some days we dismiss AI’s limitations, while other times we’re consumed by dystopian fears. This confusion comes from a limited grasp of intelligence, which could potentially hinder our evolving relationship with AI going forward.
AI raises compelling questions about intelligence. It’s worth noting that many early AI researchers were driven by their fascination with human cognition, and many current researchers are at the crossroads of computer and cognitive science too. Yet we’re a long way from fully understanding the brain, and some of the experts seem to believe that AI doesn’t need to emulate human intelligence precisely — Stuart Russell, for example, suggests AI will manifest a distinct intellectual form, independent of human cognition.
Clearly, there’s value in crafting AI that thinks more like humans. However, this doesn’t imply the resulting “intelligence” should be homogenous. Humans have their unique strengths, while AI can continue to exceed human capabilities in areas like pattern recognition. Rather than perceiving this as a threat, neurodivergence offers insights. By expanding our comprehension of human intelligence’s diverse forms, we can embrace AI as another legitimate form of intelligence and appreciate its distinct contributions, which will lead us to a more informed and constructive dialogue about our shared future.
The opportunity here lies in harnessing the strengths of both. Imagine a world where AI complements neurodivergent challenges and augments its strengths. This human-machine symbiosis could redefine intelligence in ways we’re only beginning to fathom.
Once we accept that intelligence is multifaceted, we can envision how neurodivergence could inform AI research. For instance, we could argue that AI systems might have their own neurodiversity. Different algorithms and architectures — neural networks, symbolic AI, probabilistic models — all have different “intelligences.” They excel at certain tasks while being less adept at others, which mirrors the variability we observe in human cognition and perception, and offers novel intelligence models for AI inspiration.
AI can also alleviate neurodivergent challenges, by providing personalized learning environments that accommodate different learning styles and preferences. Salesforce, for instance, developed AI that distills long texts to reduce cognitive overload for neurodivergent employees. iTherapy’s InnerVoice employs AI to bridge objects and language, assisting nonverbal individuals with a 3D avatar.
While these developments are promising, many AI initiatives focusing on neurodivergence currently operate within the paradigm of assistive technology — helping neurodivergent individuals conform to existing structures. Perhaps the next frontier involves not just fitting them into mainstream systems but recognizing and valuing the unique strengths that different neurodivergent conditions bring. For instance, some companies are starting to emphasize recruiting and training neurodivergent individuals for technical roles, spurred by findings that workers on the autism spectrum can efficiently process vast data sets and detect patterns.
An anecdote shared by a DAL colleague captures this potential: During a meeting with Japanese Paralympians using wheelchairs, it was observed that due to their unique experiences, they had become adept at providing detailed directions because oftentimes they need assistance navigating their way around the city. Their coach speculated that these individuals might excel in the field of prompt engineering due to their heightened descriptive capabilities. While disabilities and neurodivergence undoubtedly fall into distinct categories, this story suggests what the promising field of AI should do for us to truly celebrate the “diversity” part of neurodiversity.
Dr. Taylor’s inspiring journey and the experiences of many neurodivergent individuals offer a testament to the boundless forms intelligence can take. AI serves both as a mirror and a lens through which we can explore, celebrate, and augment these diverse cognitive landscapes. However, as we embrace the potential of AI, we must also remain alert and ensure it does not become a tool for marginalization. For example, we must remember that AI continues to grapple with biases, and although attention has been paid to addressing gender and racial biases in algorithms, the realm of neurodivergence remains peripheral in AI discussions. As we march towards the intertwined future, we must champion the beauty of diversity, whether it emerges from the human brain networks or the complex algorithms of a machine. Only by valuing and respecting this diversity can we truly unlock the full potential of what both humans and AI can achieve together.
DAL and DG are committed to fostering environments in which our humanistic values coexist with technology. We value the ethos of diversity and inclusiveness, and we see the embrace of neurodivergence as a direct reflection of these principles. In light of this, we recently served as the venue sponsor for the insightful neurodiversity symposium hosted by Chiba Institute of Technology (CIT). Many of our DAL colleagues facilitated sessions or attended as audience members to show our support. We believe it’s important for Japan, and the wider global community, to invest more in understanding and championing neurodiversity. Initiating dialogues such as the symposium, is a positive step in this direction. Stay with us as we explore neurodiversity and its intersection with technology and society.
Joseph Park is the Content Lead at DAL (joseph@dalab.xyz)
Edits: Janine Liberty
