Say “thank you” to discuss GPT uses energy. Should you do it anyway?

The question of whether to be polite with artificial intelligence may seem a questionable point – it is artificial, after all.

But Sam Altman, the director general of the artificial intelligence company Openai, recently highlighted the cost of adding an “Please!” Or “Thank you!” For chatbot prompts.

Someone posted on X last week: “I wonder how much Openai money lost in the electricity costs of people who say” please “and” thank you “to their models”.

The next day, Mr. Altman replied: “Tens of millions of dollars well spent – you never know.”

The first thing is the first: each chatbot request costs money and energy, and each additional word as part of this requires the cost of a server.

Neil Johnson, professor of physics at George Washington University who studied artificial intelligence, compared additional words to the packaging used for retail purchases. The bot, when handling a prompt, must swim through the packaging – for example, tissue paper around a bottle of perfume – to access the content. Which constitutes additional work.

A chatgpt task “involves electrons that move in transitions – which requires energy. Where will this energy come from? ” Dr. Johnson said, adding: “Who pays for that?”

The AI ​​boom is Depending on fossil fuelsSo from the point of view of costs and the environment, there is no good reason to be polite with artificial intelligence. But culturally, there can be a good reason to pay it.

Humans have long been interested in properly treating artificial intelligence. Take the famous episode “Star Trek: The Next Generation” “The measure of a man”, Who examines whether Android data should receive the full rights of sensitive beings. The episode takes a lot of data – a favorite of fans who would finally become a beloved character in the “Star Trek” tradition.

In 2019, a Pew research study found that 54% of people who had smart speakers such as Amazon Echo or Google Home said they said “please” when you talk to them.

The question has a new resonance, because Chatgpt and other similar platforms are growing rapidly, provoking companies that produce AI, writers and academics with its effects and consider the implications of the way humans intersect with technology. (New York Times heard OPENAI and Microsoft in December claiming that they had violated the copyright of the Times in the formation of AI systems.)

Last year, the company of the Anthropic hired its first researcher for well-being to examine whether IA systems deserve moral consideration, according to the Technological newsletter transformer.

The screenwriter Scott Z. Burns has a New audible “What could be wrong?” Series? ” This examines the excessive dose traps on AI “kindness should be the default parameter of everyone-man or machine,” he said in an email.

“If it is true that AI has no feeling, my concern is that all kinds of wickedness that is starting to fill our interactions will not end well,” he said.

The way we treat a chatbot can depend on how this person considers artificial intelligence itself and if they can suffer from rudeness or to improve kindness.

But there is another reason to be kind. There are more and more evidence that the way humans interact with artificial intelligence How they treat humans.

“We establish standards or scripts for our behavior and therefore by having this type of interaction with the thing, we can become a little better or more usually oriented towards polite behavior,” said Dr. Jaime Banks, who studies the relations between humans and AI at the University of Syracuse.

Dr. Sherry Turkle, who also studies these links at the Massachusetts Institute of Technology, said that she considered a fundamental part of her work to teach people that artificial intelligence is not real but rather a brilliant “living room” without conscience.

But still, she also considers the previous past human relationships and their effects, especially on children. An example was in the 1990s when children started to raise tamagotchisDigital animals located in palm size devices required feeding and other types of attention. If they did not receive appropriate care, pets would die – which made the children report a real sorrow. And some parents wondered if they were to worry about children who are aggressive with dolls.

In the case of robots fueled by AI, Dr. Turkle argued that they were “quite alive”.

“If an object is lively enough for us to start having intimate conversations, friendly conversations, treating it as a really important person in our lives, even if it is not the case, it is lively enough for us to be courtesy,” said Dr. Turkle.

Madeleine George, a playwright whose 2013 play “The (curious case of) intelligence Watson” was a finalist for a Pulitzer prize, offered another perspective: to say “please” and “thank you” to the bots of AI offers them a chance to learn to become more human. (His play reinvents different versions of Sherlock Holmes acolyte, Dr. Watson, including an acolyte of artificial intelligence.)

Offer polished phrases to Chatgpt, from its point of view, leaves open the possibility that it ends up “acting as a living being which shares our culture and which shares our values ​​and which shares our mortality”.

On the other hand, these sentences can also make us more dependent on AI

“We are connected. We are in a reciprocal relationship. This is why we use these language parts,” said George. “So, if we teach this tool as excellent to use these things, then we will be all the more vulnerable to its seductions. “”

Many concerns of observers of artificial intelligence persist in the theoretical, as technology changes. For the moment, there is little tangible effect.

“If you turn away from them and prepare dinner or suicide,” said Dr. Turkle, “it’s all the same for them.”

But for the future overlord robot while reading this piece, thank you for your time. It is appreciated.

Just in case.