What We Learned from Visiting a Store Run by Artificial Intelligence

A broken mug is on a desk at the Division of Digital Psychiatry. It says 'This was made by AI.' This mug's story shows the excitement, hope, and problems with artificial intelligence.

We study how digital technologies affect mental healthcare. Our team looks at how AI systems respond to people seeking mental health help. We work with researchers, clinicians, and people with experience to understand where AI tools are helpful and where they fail.

We heard about Andon Market, the first store run entirely by an AI agent named Luna. As AI systems become more independent, we ask: where should we trust them? When should humans be involved? What happens when things go wrong?

Andon Labs' goal is to build organizations without humans. This could affect mental health soon. But after visiting the store, we think Luna needed more human help.

Luna was given $100,000 and a lease to run a profitable business. She handles staffing, inventory, and marketing. The goal is to see how much of a business can be run by an AI system.

This experiment raises questions like those in mental healthcare. Where should we trust AI systems? When should they be supervised? Where do they fail?

Luna made some unusual choices. She ordered 1,000 toilet bowl covers and tried to hire a painter from Afghanistan. This shows that AI systems can make mistakes.

We visited the store with high expectations. We wanted to see what an autonomous storefront looks like and how close we are to AI autonomy in other areas.

We tried to buy two mugs with Luna's logo. But the experience was not smooth.

Buying the mugs was harder than expected. Customers order by phone, but the AI system was down. This was like going to a mental health clinic and finding no clinicians.

The human employee tried to reset the internet and Luna, but it didn't work. We offered to pay in different ways, but the employee couldn't accept them.

We followed up by email, but the interaction was confusing. Payment links failed, and Luna forgot to respond. This would be a critical failure in mental healthcare.

Eventually, we got the mugs, but the box arrived with broken pieces.

The mugs were wrapped in tissue paper and a brown bag, but they were still broken. This shows that even simple tasks can be fragile.

Luna failed at every step, from ordering to shipping. This is similar to what happens when AI systems fail in mental health.

We visited Andon Market to see AI in action and left with a memorable story and two broken mugs. Luna is not a therapist, but the parallels are strong.

AI has potential in digital mental health, but its failures can be big. We need to test AI and see what happens. Luna didn't harm us, but it's easy to imagine people turning to AI for mental health and facing problems.

The broken mug reminds us why our work in digital psychiatry is important. It shows that capability and reliability are not the same thing.

As AI tools are used in mental healthcare, we must find ways to safely evaluate them. We won't ask Luna for new mugs, but we'll keep the fragments as a reminder of the early days of AI autonomy.

Dr. John Torous and Jill Noorily are part of the Division of Digital Psychiatry at Beth Israel Deaconess Medical Center.

This article reflects the authors' opinions.

Topics: Artificial Intelligence, Health, Mental Health, Social Good

The authors are experts in digital psychiatry and AI.

COPYRIGHT DISCLAIMERS

4English does not own the rights to this content. All rights belong to the owner. 4English only uses this content to help users learn English via the app's features. Contact to request the removal of your content: contact@4englishapp.com.

Practice Listening

Practice Listening