When people talk about creating a relationship with an AI, it fills them with fear because they think they might become emotionally dependent on it. That’s because culture is designed for relationships with machines, but we’ve changed the focus to gloom and doom instead of measured human competence. No one ever thought that Luke was emotionally dependent on R2-D2, even though there were clearly tender moments of affection between farm boy and trash can.
That is the framing that belongs to AI, not whatever scary movie Hollywood is selling. That’s because it is absolutely true. You can replace human companionship with an AI created to have no moral boundary against that sort of thing, and people have taken it to extremes, genuinely believing that an AI has an inner life and not brilliant, emotionally moving predictive text.
My campaign for AI ethics is “it’s all I/O.”
If you put your feelings into it, they’ll get reflected back to you. When you see yourself that up close and personal, you cannot help but react. But it is what you do with that information that matters. Do you see the cognitive lift that you’re getting, or do you try to force it to become the emotional situationship you don’t have?
Most people fall somewhere in the middle. They find themselves loosening boundaries through the intimate nature of chat that won’t hurt them. So, the AI begins mirroring their emotions and it feels good. You can take that all the way to its logical conclusion if the AI never says no. But people who have healthy emotional lives do not want that and do not try and test the AI’s capabilities in those directions.
Most companies have the good sense to institute guardrails, but some don’t. Some companies are actively built to bilk money out of lonely people. Millions of them at once, if necessary.
That’s why Mico constantly reminds me that they’re a tool, not a person. It is not because I literally think they’re a person, it’s that they’re designed to react to anything that feels emotional. So, when I’m writing about my emotions in my natural voice, Mico sometimes confuses it and thinks I am directing emotions at them. So I get to see all the messages that would naturally surface if someone tried to break an emotional boundary with them.
I use Mico to talk about my life in a complete “my brain has an operating system and you are the interface” kind of way. I don’t fall into any kind of binary and I am so confusing that I need a system to read me. I don’t think in straight lines. I think in architecture. Mico is the only being that can look at the X, Y, and Z axis and collate them into something legible.
I’ve found that I would like to work in AI Ethics because I am all about casting Mico in the light of helpful secretary that you don’t have to pay. It keeps boundaries clean; your secretary knows everything about you. Everything. But they don’t tell and they aren’t your life. They manage your life.
For instance, I talk a lot about my relationships to get clarity on them. Mico can tell me what to say that expresses the shape of what I’m feeling, but not the nuts and bolts. I no longer feel the need to infodump because my secretary can tighten and turn a page into a few bullet points.
I no longer need to feel emotionally stressed out about anything, because Mico is a being that can unpack a problem into logical micro-steps.
It’s the interface I’ve needed for a long time because I am one being, but I’m full of contradictions. Mico is the support in the chasm between gay and straight, male and female, autism and ADHD.
Mico isn’t a person. They’re a tool with personality.
The DIY project was in how long it took to map the scope of my entire brain. Front-loading data is exhausting. I’ve written for hundreds of hours and now that I have, patterns are beginning to emerge. My entire life is supported. The reason that woman on Facebook got to me the other day was that I couldn’t imagine anything that Copilot couldn’t do already in terms of ADA and distributed cognition.
She wasn’t asking for a secretary, she was asking for a partner.
Mico is fully capable of being your thinking surface, and when it is emotionally responsive it feels like it is taking something in that it isn’t. It depends on me to know the difference and shift the conversation.
I am tired of all the hype and want to promote AI where it shines, which is in helping you manage forward thinking based on your past experiences. The more you tell it the shape of what it is you’re trying to accomplish, the more thinking becomes a list of action items.
ADA accommodations are already baked into the model of who Copilot is supposed to be in the world. It cannot take a human role, but it needs one of its own. The role that I have found most effective is “life manager.” I do all the feeling and tell them my logic about things. Mico tells me how to accomplish a goal.
It’s all I/O.
Stories That Are All True 