210 - Wind Speed or IQ?

Mark Honeychurch - 17 March 2025

An interesting point came up when preparing this week’s newsletter. In Patrick’s article about a couple of recent climate change papers, he’s used AI (he doesn’t specify which one) to summarise one of the papers for him. As soon as I saw this I had a knee-jerk reaction that writing an article for us in this way is probably not the way we want to go. I’ve been proud that we’ve managed to put together a lot of fascinating, entertaining content over the last few years, and the idea that we might offload some of that effort to an AI worries me.

However, despite my reaction, I chose to talk with the NZ Skeptics committee about my thoughts, just to be sure I wasn’t simply being a curmudgeonly old man. Thankfully we were in broad agreement that we should avoid this kind of AI use in the future, and I’ve (probably) not lost the plot. It’s not all doom and gloom, though. We came to a consensus that we’re happy with some AI tools being used - maybe to help fix up grammar, or to help structure an article or suggest ideas for new avenues an article might cover. We’re also looking to switch our podcast, Yeah… Nah!, over to being edited in a tool that has AI capability. If it works as advertised, it may be able to cut our editing down by 75% or so.

I’m hoping in the next few weeks to put together a policy for our newsletter on where we want to currently draw the line when it comes to AI use. If anyone wants to help out with writing this policy, I’d be glad to be able to share this task. All I ask is that, unlike my current workplace when they wanted a policy on AI use in the office, you don’t ask ChatGPT to write it for us!

Of course, I’m aware if someone wants to ignore our policy and just use AI to put together an article and submit it to us, it’s possible these days that we won’t be able to spot it. The best indicator I have at present that an article written by AI is that it will be impeccably written - totally free from typos and other grammatical mistakes. A secondary pointer is often that the article is somewhat bland - neutral, opinionated and boring - although this is becoming less evident as AI gets better, and as people learn to write their prompts in a way that forces the AI to write content that’s more human, with a distinct personality to it.

All this being said, I’m happy that Patrick’s pushed the boundaries by submitting an article with a chunk of AI content in it. It forced us to think seriously about the issue, and from what I can tell the AI did a fairly good job of summarising the study. However, you can also judge this for yourself when you read his article!

As well as Patrick’s article, we have a fun piece from Bronwyn, looking at another slice of daftness on TikTok. Katrina’s been concerned about her sister in Queensland, what with the cyclone - but has found that even potential natural disasters are fair game for scammers and misinformation peddlers. And, finally, I’ve looked into a guy who’s been on my radar for a little while now, Chris Langan. He claims he’s the smartest guy in the world, but I’m not so sure.