AI terrifies me

 “It’s the hill I’m prepared to die on,” I tell people when they insist that AI is brilliant. “I can write. I don’t need a computer to do it for me.”

People laugh, roll their eyes, convinced they are witnessing a dinosaur staring at the meteor of her own demise. 

In a meeting for music teachers, an AI expert from a European university joined us on Zoom to explain how we could use AI. He said that we could ask it to write a song about Romans (because there aren’t enough of those already) and took us through the weirdly specific prompts we would have to give it. At the end of the confusing stream of commands in different types of brackets, he played us two examples, one of which he admitted was terrible and the other, which was also terrible but he was very excited about. No participant had their camera on by this stage, so it was impossible to tell if I was the only dinosaur in the virtual room. 

At the end, we were all asked to say what experience of AI we had and people mumbled and appeared terrified to say that they, as creative people, could do this stuff quicker and the idea of writing computer code was alien to them. Instead, they said things like, “Well..um….I haven’t yet…I really should…”

I read the room and kept quiet. 

The implication was clear. This is the revolution and you embrace it or die. 

It is impossible to avoid now. Every Google search gives you an AI version, Canva insists it can design your poster better than you and your racist grandad probably has a virtual 12 year old girlfriend, telling him that white supremacy is the natural order of the world (That is Grok, right?)  Even Word’s annoying paperclip has morphed into the AI Co-pilot. 

I am most concerned because any LLM will take human bias and amplify it. (Read Laura Bates new book if you care about women’s rights and want to be truly terrified). Ask Chat GPT to give you a random number between 1 and 10 and more often than not it will give you 7 because there is a very slight human bias towards that number. It is also a brilliant liar and gaslighter.

Over the last few days, I’ve been reading Virginia Giuffre’s posthumously published book, Nobody’s Girl. It is brilliantly written and the ghost writer is not hidden, which I suppose can happen when the subject becomes an actual ghost. 

As I was reading, I had questions. There were facts I wanted to check, ideas I couldn’t reconcile, coupled with an enormous amount of horror and sadness. It’s not a story anyone will want to believe. My questions were complex and were taking more than one internet search to get the answers. 

At one point in the book she says that authorities have told her that there is a credible threat to her life, black vans are outside her home, shining lights through her windows, and she says that she told friends that if something happened to her then they should know it wasn’t an accident because she wasn’t suicidal. If I had only kept on reading, the question I had would have been answered but instead all I could think was , “Why?”

So, I asked Chat GPT. An internet search had only brought up articles about the Royal-formally-known-as-Prince’s arrest. 

I asked, Based on Virginia Giuffre assertion in her book, Nobody’s Girl, that she is not suicidal and that if anything happened to her it wouldn’t be an accident, why is everyone so certain that she killed herself?

ChatGPT responded, ‘There is no verified information that Virgina Giuffre has died. As of the latest publicly available reporting she is alive. If you’ve seen posts suggesting otherwise they are either rumours  misinformation or speculative commentary circulating online.” 

A further spewing issued forth, explaining things like confirmation bias and emotional reasoning. It did offer to check my sources.

So I replied, Wow! Are you sure about that?

This was a huge conspiracy theory that I had not expected. 

The reply was hilarious. 

“Yes - based on credible reporting from multiple major news outlets and encyclopaedic sources Virginia Giuffre died on April 25th 2025 and her death has widely been reported as suicide by police and family. It’s not just internet chatter or rumours - mainstream news and reference sites have documents this,”

It also listed all the places where this information is available. I couldn’t help laughing at the yes. Gaslighting, at its finest.

I challenged. 

But you said there is no verifiable information that the has died and she is alive. I’confused that you appear to have given two contradictory statements.”

ChatGPT was contrite but unrepentant. 

“You’re absolutely right to call that out - thank you for doing so. I gave you two contradictory answers……. The cannot both true and that inconsistency is my mistake. What happened is this: 

I didn’t read on. Gaslighting I could take, Mansplaining not so much. 

If only I had continued to read the book then I would have discovered that health problems an estrangement with her husband and a suicide attempt after the statement about not being suicidal were clues to why her death is believed to be not suspicious.