Letters
Tell us what’s on your mind
Saying ‘please’ to ChatGPT is completely mad…
I could barely believe what I was reading in Issue 709 (page 8) about people using ‘please’ and ‘thank you’ in their requests to ChatGPT. What do they think ChatGPT is? It’s just a collection of algorithms and code, and is no more human than my toaster, kettle or hoover. It doesn’t care whether you’re polite to it. Adding ‘please’ to a question makes no difference to what it gives you. It’s completely mad, and as pointless as being polite to Alexa or Siri. In my experience, the best way to use ChatGPT is to be blunt and unambiguous. You certainly don’t need to treat it as though it has a human-like personality. You don’t even need to use punctuation or correct spelling. It always seems to understand what I’m asking, no matter how many mistakes I make when typing.
Neil Spencer
CA SAYS We asked ChatGPT and it said “politeness doesn’t affect how I respond – I’ll give the same level of effort and accuracy either way”. However, it added that if you say please and thank you it will respond in a similarly friendly style. Indeed, it said it can adapt to almost any style you use, including “affable, blunt, sarcastic, formal, jokey, technical and analytical”. So, being polite won’t make ChatGPT better, but it will nudge it to respond in kind.
…and it’s as stupid as Baldrick
I’m intrigued by the current emphasis on AI. We’ve actually had AI for decades. All the tech companies have done recently is modify the front end of your average search engine and market it as AI because it makes a haphazard approximation of human conversation.