- Rishi Sunak said his daughters say please and thank you to AI chatbots despite advice not to.
- Sunak noted that politeness to AI wastes computing power but his daughters insist on being polite.
- His daughters believe being polite to AI is important if AI takes over the world.
Rishi Sunak, the former Prime Minister of the United Kingdom, told NDTV that his daughters say 'Please' and 'Thank you' while talking to artificial intelligence (AI) chatbots. Speaking to NDTV CEO and Editor-in-Chief Rahul Kanwal at the NDTV AI Summit, the former British PM noted that, though using gestures of gratitude while talking to AI wastes millions of dollars in computing power, his daughters do so anyway.
"My two daughters are very polite when they use these models and chatbots. They always say 'please' and 'thank you' during conversations. I told them, 'You don't need to say "please" and "thank you." This isn't a human. It requires a lot of computer power, so it's best not to use those words," Sunak said.
He then shared his daughters' response to his suggestion, who told him that "if AI takes over the world, we'll want to be polite to AI."
"I thought it was a good system," Sunak said, adding, "I think it shows how important it is to make people trust technology."
Rishi Sunak and his wife, Akshata Murthy, have two daughters – Krishna and Anoushka.
Should We Say 'Please' And 'Thanks' To AI?
According to OpenAI founder Sam Altman, saying "please" and "thank you" to ChatGPT and other AI models is wasting millions of dollars in computing power.
AI depends on large data centres built around high-density computing infrastructure. These facilities are built on acres of land, use a substantial amount of electricity, and require continuous cooling. Adding extra words to prompts adds to operating costs at the scale of billions of prompts.
However, Google DeepMind principal scientist Murray Shanahan has a different opinion on the matter. Shanahan believes that saying 'please' or 'thank you' to an AI can actually make it more productive.
Speaking in the Google DeepMind podcast in 2025, he said, "You can say thank you. Yeah, there's a good reason, good scientific reason, why that might get better performance. Say it's role-playing a very smart intern, right? Then it's going to just role-play, maybe being a bit more stroppy if they're not being treated politely. "
