I was out all day yesterday, so missed the WordPress prompt. Of all the total nonsense questions they usually ask, which seldom prompt me to write anything, I could have offered a plausible response to this one.

So thanks, Lou. Their question totally passed me by until I read your response this morning.

The question was:

What do you think is the future of AI?

In terms of blogging, I’m the same as Lou. I use it to generate images. I can’t really imagine that changing, ever, my goal is to have an enjoyable way of spending my time (which happens to be writing). I don’t want AI to crank out a hundred pages per day for me, for example, so what’s the point if it can? In other words, it’s the process which is important, not the end result. Everyone I know on this platform seems to be the same.

When I ask AI to generate things, the thing that infuriates me about it is that when I have an image in mind, I have to word the description so precisely before I can generate something which even approximates to my idea.

I think that shows the limitations of AI. Having to go to the n^th degree before we get the output we want.

It also shows where the future job opportunities will be. They’ll be for people who are proficient in communicating with AI tools. You know, to get the desired output first time.

Which, really, is the same challenge that there has always been. For years, I could make a computer do something useful, because I happened to know specific programming languages that a computer, ultimately, could understand. I reckon this’ll be the same, except in the future, I’ll be speaking in English, and having to use certain words, rather than having to learn a whole new language.

So in that respect, I don’t think the problem has particularly changed with AI. Sure, the tools have changed, but that’s just evolution. The actual problem – communicating with a machine so it does something useful – is the same.

Where I do find AI incredibly useful, though, is when I write. One of the tricks is to make our stories as plausible as possible, right? And in that regard, there are all sorts of tiny facts that we don’t know, that we have to go and research, in order to make a story plausible? AI can now make that research instantaneous.

For that reason, I use my AI program (Copilot) pretty much every day. Examples of recent questions to it are:

• What would army slang be for a Field Marshal? Because I have no knowledge of army life, and one of the characters in my current novel is a senior officer. How would he refer to his boss? In fact, the AI said there isn’t a more specific term, so I just ended up referring to him as “the top dog”.
• What were the winning lottery numbers on a specific date, 25 years ago? Because my novel is set 25 years ago, and I’m trying to be vaguely realistic.
• Who won the UK Golf Open in 2002, and what were the starting odds? Same reason. If you’re interested, it was Ernie Els and I think his starting odds were about 20-1. I remember thinking that those were quite long odds, since he was also described as “one of the favourites”, but apparently that’s typical in golf. I guess there are just so many people in every tournament that nobody is ever a sure thing.
• Can you freeze cheese? Nothing to do with the novel, but the kitchen is a foreign land to me! And, of course you bloody can 🤣.
• and so on.

But do you see, they’re all just tiny questions which you might bounce off an assistant, say. Or you might find out in a library. None of this is particularly “intelligence”, it’s just time-saving to get the answer immediately.

So I pretty much see the main benefit of AI as being my PA.

And I guess the big danger of AI is therefore relying on it too much. For example, if we had a human PA, we might dismissively say, “can you arrange my trip to Melbourne, please?”

Could we really say that to something AI? I mean, in theory, yes. It could make a plan. But there are so many smaller tasks in that instruction – book a taxi to get to the airport, book a flight, cater for delays going through immigration, book a hotel at the other side, etc. – that the human in me knows that there are a thousand little things that could go wrong, where we might have to improvise on the spot. So I’m kinda wondering how much the initial plan would resemble the actual events.