AI is changing how we think, not replacing it | Letters

Wendy Liu’s thoughtful piece on AI and cognitive sovereignty raises real concerns about labour redundancies, the hype and the environmental cost (I avoid AI tools because thinking is supposed to be hard. It’s what makes us human, 24 May). But I think she allows those legitimate grievances to colour a separate and more interesting question: what is AI actually doing to the way we think?

I use AI heavily and it has changed how I think, but not in the way she fears. It has made me more curious, not less. I now ask questions that I wouldn’t have known to ask and explore territory I would never have had time to reach. Yes, I offload research, but that offloading doesn’t empty my mind, it frees it.

Liu worries that we are sacrificing depth for convenience. My experience is the opposite: the breadth AI gives me has led me to deeper questions, not shallower ones. Just as books didn’t make us lazy readers and Google didn’t stop us researching, AI is changing the shape of thinking rather than replacing it. That has always felt threatening in the moment and obvious in retrospect.

What strikes me most is the optimism she leaves out. For most of human history, serious intellectual inquiry was the preserve of the privileged few. AI is changing that. Millions of people can now bring genuine curiosity to hard problems, and we should encourage that. That seems to me like exactly the kind of development we should be fighting for, not against.

I guess it remains to be seen if AI will let us explore this as we want to and not how big tech wants us to do it. Maybe Liu can write about that next, as it is a worry.Richard ThackeraySheffield

I’m in full agreement with Wendy Liu’s views on AI. As a (pre-AI) software engineer, I created some very successful products through hard work and thinking through every idea. I can’t imagine releasing software generated by AI and having no clear understanding of its components. I wouldn’t like to know if the aeroplane I’m on is controlled by code untouched by human intelligence.

As a musician, I see a parallel with tablature. I have always avoided the piano-roll technique to learn guitar parts. I don’t want to play what, for example, Mississippi John Hurt played note for note even though it would sound good. I would much rather listen to Hurt and learn the songs by ear, interpreting them on the way. So my versions aren’t exactly “right”, but they’re mine and I can improvise on the tunes since they’re not just muscle memory. In fact, Hurt rarely played tunes the same way twice.

As Liu says, thinking is good. Maybe a piece that is self-created is a bit wonky, has malapropisms, unusual punctuation, bum notes and so on, but at least it’s human, and I believe has embedded emotion that readers can feel, maybe subliminally, as opposed to perfectly polished AI output. Phil SnellOtley, West Yorkshire