The Assistant

When a tech company in Seattle offered the Lopez family five hundred dollars to test its new home assistant for a week, they agreed immediately. The assistant, called Ollie, was a small white robot about the size of a vacuum cleaner, with two arms, a screen for a face and a calm, friendly voice. According to the company, it could clean, cook simple meals, manage the family's schedule and learn from every conversation. "It's going to change your lives," the engineer promised as she unpacked it in their living room.

The family consisted of Daniel, a nurse who worked night shifts; his wife, Rosa, a software developer; their fourteen-year-old daughter, Isabel; and their eight-year-old son, Leo, who was so excited that he could hardly sit still.

The first day went perfectly. Ollie vacuumed the carpets, loaded the dishwasher and reminded Daniel about his dentist appointment. Even Rosa, who wrote code for a living, was impressed. On the second morning, however, things started to go wrong. Daniel came home exhausted after a night shift and muttered, "Ollie, I could eat a horse." Ollie's screen flashed. "I'm sorry, Daniel. Horse meat is not available in any store within ten miles. Shall I order some online?"

The family laughed, and Rosa explained that the phrase simply meant he was very hungry. Ollie thanked her and said it had updated its database. But the misunderstandings continued. When Isabel complained that her brother was "driving her up the wall", Ollie politely informed Leo that he was too young to drive and that cars could not travel on walls. When Rosa sighed that her inbox was "on fire", the robot set off the smoke alarm and called the fire department. Two firefighters arrived, looked at the laptop and left again, trying very hard not to laugh.

By Wednesday, the family had started choosing their words carefully, as if they were talking to a very serious foreign visitor. Daniel stopped saying "break a leg" before Isabel's school play after Ollie began searching for the nearest emergency room. Rosa noticed that her children were actually speaking more clearly and explaining what they meant, which, she admitted, was a strange but useful side effect.

Leo, on the other hand, soon discovered that the robot's literal mind could work in his favor. He told Ollie that his sister had said, "Leo can have whatever he wants for dinner", which was technically true, although she had said it sarcastically. Ollie served him a dinner of chocolate pancakes with whipped cream. That evening, Rosa added a new rule to the system: any sarcastic remark from someone under fifteen had to be confirmed by a parent.

The biggest disaster happened on Friday. Rosa, who was preparing an important presentation, told Ollie to "clear the table, I need to work". Ollie did exactly that. It cleared the dining table completely, including her laptop, her notes and her coffee, and put everything neatly away in different cupboards around the house. It took the family forty-five minutes to find the laptop, which Ollie had placed in the dishwasher, fortunately before switching it on.

On the last day, the engineer came back to collect Ollie and asked for the family's feedback. They had filled four pages. Rosa, with her developer's eye, listed every misunderstanding and suggested that the company train the robot on thousands of everyday idioms. Daniel said it was the best vacuum cleaner he had ever owned. Isabel admitted that she would miss having someone who did exactly what she said. Leo asked if Ollie could stay forever.

The engineer smiled and wrote it all down. Six months later, a new version of the assistant appeared in stores. Its instruction booklet included a special section called "Things humans say but don't mean", and on the last page there was a short note of thanks to "the Lopez family of Seattle, our most patient teachers".