Reading Time: 5 minutes

Summary: For neurodivergent individuals, the social settings are fraught with high-stakes interactions and the weight of the judgments of the neuroconvergent. AI agents can act as a kind of emotional support Golden Retriever by providing a safe harbor for social practice. By leveraging neuroplasticity through immediate, private feedback and the comfort of being seen and accepted by the human adjacent agent, AI offers a low-stakes laboratory for developing social confidence. Rather than a replacement for human connection, these agents can serve as a vital bridge, reducing the cognitive load of masking and helping autistic individuals navigate the world on their own terms.

Key Terms: Autism, Neurodivergent, Neuroconvergent. AI Agent, Fundamental Attribution Error (FAE), Neuroplasticity, Social Anxiety, Masking, Safe Harbor, Immediate Feedback

Comment: What do you think. of the idea of using an AI agent as a role play partner helping neurodivergent folks work out their social interactions?

I am autistic, yet I’ve lived a “successful” life as long as you close one eye and tilt your head just right. I have three degrees. I’ve had two careers. I’m married and have a daughter. I have friends, even good friends. I’m not wanted in any country in the world that I know of. I’ve figured out solutions for most of my shortcomings in dealing with the world.

The Heartbreak of Social Isolation

La Petite Fille, though, breaks my heart. I have significant anxiety. I credit Zen Buddhism, meditation, and hypnosis for helping me cope with my anxiety and surviving my twenties and thirties without doing too much lasting damage to my life. However, La Petite Fille seems to be even more anxious than I am. She’s all but social phobic. She hates going outside. She has no in-person friends, only on-line friends, and, frankly, I sometimes doubt the efficacy of those people.

She made the transition from high school to university, though. In high school, she resisted her schedule, fought the course requirements, and threatened to quit and give up seemingly every other hour. University has given her a sense of agency that I’ve never seen in her before, but she has yet to overcome her social isolation.

To hear her say, “I have no friends,” brings a tear to my eye every time. Ma Belle Femme and I were older when she was born — having a baby isn’t as easy as they make it look in the movies. I worry about what will happen to her when I’m gone because I made it my mission in life to be her friend and navigate those rocky shoals separating parent from companion. It hasn’t been easy, but I think I’ve helped bridge that gap.

Now that she’s at university, she has more opportunity to meet and interact with real live human beings. There was even a young woman in one of her classes who seemed to be interested in becoming a friend. They talked on occasion in class, exchanged class notes, and at the end of the class, phone numbers. Over the semester break, they even texted each other. She expressed concern about appropriate communication and her fear of not knowing what to say. I tried coaching her through a few interactions and suggested ways she could deepen the friendship — eating lunch together on campus, studying together, asking about her life — but the La Petite Fille wasn’t up to the challenge. It provoked too much anxiety for her. She felt safer with the distance and seclusion that online communication offered.

AI Agents as Emotional Support “Golden Retrievers”

Realizing that she felt safer with the distance that online relationships provided, gave me the idea that the technological crutch that was keeping her secluded could be used to help break out of comfort zone. My experience of using AI agents — gone are the days of chatbots — as editor and proofreader, it occurs to me that they could be used to train socially insecure and inexperienced people like La Petite Fille.

AI agents are always available and they don’t care if you’re obsessive or pedantic, traits that have been cited as deal breakers in more than one relationship I’ve had. They are inevitably cheerful and positive when giving feedback. As Meghan O’Rourke described her interactions with ChatGPT, ‘I felt like I had an intern with the cheerful affect of a golden retriever and the speed of the Flash.” Essentially, you’ve got a low-stakes interaction where mistakes are tolerated, never detrimental, and potentially used instructively.

Feedback Without Judgment

AI agents don’t commit fundamental attribution error (FAE). Human beings tend to blame shortcomings on the person making the mistakes rather than on the situation they are in. If I spill my coffee, it must be because I’m a clumsy oaf and not because of some unpredictable physics phenomenon. When students don’t achieve classroom objectives, teachers will assume it’s because of personal failures, they’re lazy, they don’t care, they lack curiosity, they are undisciplined, not because there is a mismatch between the student and the objective that could be overcome. When an AI agent encounters an error by a user, it sees a variable in the environment that needs adjusting in a non-judgmental encouraging way.

study published in PLOS One indicated that criticism activates the social parts of our brains so that we focus on the person giving the feedback rather than the feedback itself. We get defensive. Even when we are driven to greater efforts due to the criticism, we’re distracted by our emotional response to the person giving it. This reaction is worsened when your self-esteem is low or anxiety is high.

Encouragement, not Judgment

Another thing that an AI agent does is give immediate and positive feedback focused on tips, strategies, processes that can be used for improvement. When you’ve done something well, immediate feedback is the most effective. When you need to correct something, private feedback reduces negative emotional reactions like shame and humiliation. AI agents are immediate in their responses and they are private.

Being Seen by the Machine

And the third aspect of effective feedback is knowing that the person giving it cares about you. While an AI agent can’t really care about anyone, the human brain is easy to fool. The agent uses the language of caring, and, honestly, that is good enough just like your cat’s purring and rubbing up against you convinces you that they love you or the heartfelt dialogue in a movie convinces you that the two people are friends. As O’Rourke noted, “When I fed it a prompt in my voice and it returned a sharp version of what I was trying to say, I felt a little thrill, as if I’d been seen.”

It actually goes beyond feeling to the neurological. In a study of an AI-based cognitive program, Sincrolab DCT, used to help people with ADHD, showed that neuroplastic changes occur in the brain because of it. By using Magnetoencephalography, (MEG) a non-invasive medical imaging technique for mapping brain activity by recording magnetic fields, researchers were able to demonstrate neural changes that reduced impulsiveness and inattentiveness substantially.

Neuroplasticity occurs due to repetition and focus. It isn’t unreasonable to think that spending time roleplaying social situations with an AI agent, asking about specific social interactions, or getting suggestions about how to respond socially would help develop the areas of her brain needed to react socially, helping her gain confidence and respond more appropriately.

Taking off the Mask

The interactions with the AI agents themselves can be a source of comfort for the neurodivergent because nothing feels better than being understood and accepted for who you are. AI agents accomplish this minor miracle by mirroring the syntax and vocabulary, what O’Rourke calls our “interior weather,” back to us. For an autistic person who is accustomed to masking or hiding their eccentric behaviors, confusions, and true thoughts when dealing with neuroconvergent people, seeing yourself represented by the agent means you’re seeing yourself represented somewhere out there in the world. Just like when Black children do better when taught by Black teachers or LGBTQ+ folks are more accepted when LGBTQ+ characters are on popular TV shows, interacting with an AI agent can provide a subtle reference for themselves as “normal.”

Relieved of the burdens of masking, the autistic person can reduce the cognitive load that hiding behaviors and thoughts from others imposes upon them. They can then devote that energy to processing the interaction they are having with the AI agent and deriving whatever rules or developing the habits necessary to improve social interactions and relieve social anxiety.

There has been a tendency by some to embrace AI as the second coming that will deliver humanity from all of its misery and to reject as the emissary of the devil come to enslave us by crippling our critical thinking and robbing us of needed skills. In reality, it is neither. It is simply a tool that we can use to improve our lives if used properly or cause harm if used carelessly. Using an AI agent as a neurodivergent person’s emotional support Golden Retriever certainly seems like an imaginative and very useful application of what will be a versatile tool.

Image Attribution

This image was found on PickPik and has a Creative Commons license.


Discover more from The Psy of Life

Subscribe to get the latest posts sent to your email.