– We like products better if we can forgive them for their mistakes – NRK Østfold – Local News, TV and Radio

“We knew what we were going to call it before we bought it,” says Merith Berg.

The robot stickers were purchased before the vacuum cleaner became part of the family.

The robot is named “Toothless” after the toothless dragon in the movie Dragon Trainer.

– It's fun to use a robot vacuum cleaner with eyes that eat up all the dirt on the floor, Berg says.

Run the vacuum several times a week. Sometimes he finds it difficult to get around the house. Berg then moves furniture until “Toothless” comes on.

It is not unusual for people to adapt their homes to suit robots. You may also be doing it more than you know.

This is what Hanna Williams Sollund says, who together with Leif Armut Brastad wrote a master's thesis on humanizing robots.

“You want to make the task as easy as possible for them and you don't want them to struggle,” says Sollund.

“We also feel that the robot vacuum cleaner takes on a personality when we give it a name,” she says.

Toothless is just one of many vacuum cleaners found in Norwegian homes, named after it. Here is a group of electronic assistants:

Humanization of robots

When we give a vacuum cleaner a name, it means its design works, says artificial intelligence expert Inga Strömke.

She explains that technology companies want us to empathize with robots.

Berg feels no sympathy for Toothless.

But she often tells it to return to its charging station, even if the robot doesn't understand what you're saying.

For us humans, it's completely natural: we humanize things, explains psychologist Sven Øverland.

-This doesn't just apply to robots, it applies to everything from pets to spiders. If you have a cup and you draw two dots and a line, people see a face.

An extension of us

Technology companies are taking advantage of this.

– This is done in consultation. We like products better if you're willing to thank them or forgive them for their mistakes, says Strömke.

The strategy can also be seen in the names of well-known language models, such as Microsoft's “Copilot” or Snapchat's “My AI” language model.

– They should be a part of you and an extension of you, says University of Oslo professor W Scientif Peter By Brandtzig.

This week, OpenAI, which is behind ChatGPT, released its latest version.

GPT-4o responds much faster than previous models. The company itself writes that the new version is Get closer to natural conversations with robots.

According to OpenAI, the new model can pick up on your mood and adapt to it when it verbally responds to what you're talking about.

Polite to robots

Psychologist Overland adopted several linguistic models and decided to be polite to machines.

– Robots will not be harmed if we behave badly towards them. But I believe that the more rude we are to the things around us, the easier it will affect us as humans.

But robots do the job, at least in some cases.

Whether machines are affected by the way we speak depends on whether the machines actually learn as they use them, says AI expert Strömke.

Developers can give models different abilities. The robot vacuum probably won't learn from our behavior, nor will it care how we interact with it.

There probably won't be well-known language models like ChatGPT either, Strömke explains.

He can learn from us

But technology that can learn and be influenced by us exists.

“We don’t know whether this possibility has been activated or not,” says Strömke.

This type of artificial intelligence is considered unpredictable by large companies, explains Maria Slavkovic, an artificial intelligence researcher at the University of Bergen.

“We know that it is extremely dangerous to allow models, or AI systems, to learn completely without human supervision,” says Strömke.

– Because he has Billions of users. It is not good to say whether it will be withdrawn In a good direction or not.

Language models, such as ChatGPT, are built on large amounts of data.

It calculates the probability of each word and guesses the answer.

For a long time.