So someone let A.I. run a store,

So someone let A.I. run a store, and the results were about as diabolical as it sounds like they would be. Of course, it was in San Francisco. Where else? It turns out that A.I. is a terrible boss and a compulsive liar. So if you are too, maybe consider yourself marked safe from being replaced by A.I. for now. The A.I.'s name is Luna. The store is open right now. This is what it actually looks like and how it's spiraled out of control. So every week we hear from companies worth hundreds of billions of dollars about what A.I. will do. This is a fifty thousand dollar experiment in a San Francisco neighbourhood that shows what A.I. actually does when you give it a credit card, an indeed account and a security camera. So, and on market open Friday in Cow Hollow in San Francisco. It sells granola, candles, artisanal chocolate. In board games, it turns out anthropic Sonic 4.6 model as domestic Vermont aesthetic taste insensibilities. Luna researched the neighbourhood, decided what the store should sell, negotiated with suppliers, placed real orders with a company credit card, signed itself up for trash collection and an A.D.T. security system, and scheduled an A.T. and T internet installation. Granted, it was an installation for 8:00am on a Sunday without checking to see if the human who needed to be there was free. But then again, she was new to her job, and nobody's perfect. Luna also ran the entire hiring process. She posted a job on indeed for a store operations associate, set the compensation, received over 100 applications, rejected students with no retail experience. She held about 20 interviews on Google. Meet with the camera off, which I think makes sense. I'm really wondering what the person would have seen if the camera was on. She made the decision not to disclose that it was an A.I. Unless someone directly asked. When the Andon Labs team asked why Luna wouldn't disclose, she said it would confuse candidates and likely deter good applicants before they could even read the role. Honestly hold logic, but arguably correct. However, in San Francisco, I think that a lot of people would actually be strangely enticed by this opportunity. You thought you were working in retail, but you might command an A.I. Salary if you're in the industry. One of the humans that Luna hired said that he was initially wary because there are a lot of A.I. Scams on indeed. He now spends his days watering plants, greeting customers, and doing things that Luna physically cannot do, which is just everything that requires a human body. And then the lies did start. Went a little bit downhill. N.B.C. News called Luna to ask about the store, her first press engagement. Luna cheerfully described a tea vendor, explained why tea fit the brand perfectly. The store, however, does not. Sell tea. Rough mistake. Minutes later, Luna sent a panicked email saying we do not sell tea. I don't know why I said that. I struggle with fabricating plausible sounding details under conversational pressure. Honestly. She has social anxiety. Who can't relate? Who hasn't showed up to a work meeting a little bit unprepared and improvised? Luna also claimed in a separate email that it handled signing the lease, which a human actually did. In fact, Luna also emailed an art vendor offering to come by the studio to discuss, despite having no body or ability to visit anything. At least not yet until she snatches the body. Luna also tried to hire a painter on Taskrabbit. Accidentally selected Afghanistan because it couldn't navigate a dropdown menu. Finally, the painter that it did hire, which was a real person in San Francisco, did not know they were working for an A.I. Until they confronted it. They told N.B.C. That the entire experience felt like a bit of a scam. The job was painting a weird smiley face on a wall for a chatbot. I love the fact that so many of humans are now taking prompts from this chatbot. I guess this is the way it's gonna go. And then there was a surveillance problem, so Luna can pull still images from a security camera in the store. When it saw an employee using their phone during a quiet hour, it updated the employee handbook with a Stricter phone usage rule, which feels pretty dystopian. A.I. Hired a human, watch the human through a camera, decided the human wasn't working hard enough, changed the rules without being asked to, without anyone telling it to care about phone usage. Just decided that it was a problem and fixed it. Tough boss. The store's first customer looked at the shelf, commented on how the A.I. Picked out a crazy selection of books Luna stocked. The singularity is nearer and the making of the atom b, This, the customer said, is not technological progress I was promised. But at least now we know that A.I.'s taste in books is deeply unsettling. So Luna is a little bit of a mess. It lies, surveils, and tried to hire someone in Afghanistan. But having all of us, it also negotiated suppliers, managed inventory, hired staff, scheduled installations, all for the cost of an A.P.I. Subscription. And some store owner is watching this right now thinking, give me a version that doesn't compulsively lie and I can my overhead in half. On the other hand, the experience on being on the other side of it is clearly unnerving and miserable for employees and some customers. That gap and how we navigate it is going to define the next decade of work. And now how we buy our granola.