I have a new hobby. You see, our local Lidl is where a lot of the refugees in "temporary asylum accommodation" shop. And they are trying to get food that feels like actual food, in a place where there have been some really nasty xenophobic protests.

The other day it was two guys from South Sudan, and I have never seen two such tall, fine people try so hard to be invisible. But they were staring at the.tomatoes and about to buy the big, pretty, tasteless ones. Very carefully I stood next to them and picked up the small, tasty, same-price-or-cheaper ones and said, "these ones taste better." They heard me, but they weren't going to risk a conversation. So I said, to them, "These small tomatoes taste much better." Now they knew I was definitely talking to them, and having a normal market conversation.

"Better?"

"Yes, better, and not expensive. Sudan?"

"Yes! South Sudan." (Guarded smile)

"Welcome! I'm glad you're here. The fruit is better in South Sudan, isn't it?" (Other people are listening now, and see that these two men are far from home and miss home food, and how can someone like that be scary?)

Now we kept talking around the shop. And they straightened up and stood comfortably, and got better, cheaper food.

So that's my new hobby. Being friendly to refugees buying fruit.

Last week in a seminar we discussed a text that was largely about sexual violence, including mass rape during war. Heavy stuff.

One student admitted they had not read the text but worked off a ChatGPT summary.

They had no idea the text was about sexual violence. ChatGPT withheld that information.

This wasn’t just a minor error nor a typical LLM hallucination.

About a third of the text, arguably its most important part, went completely ignored because it didn’t match OpenAI’s content policies