How I use AI

Seven apps that not only improve my productivity, but allow me to explore the world, old and new

I sometimes forget that I have one of the best jobs in the world. I am paid to explore. Three centuries ago, exploration involved a sturdy horse and a willingness to risk scurvy. A century ago, it meant trains, steamships, and reporting via unreliable telegrams. Today, equipped with robot guides, I can explore new worlds without ever leaving my desk.

Midjourney

Or I can simply create them. Midjourney, for those who haven’t tried it yet, is an artificial intelligence tool that transforms words into images. As regular readers of this blog will know, I use Midjourney to create images for each post. I love it. It gives me the creative outlet I have always craved, and it has a surprising influence on my writing as well. Creativity turns out to be multiplicative. The images I make often spark new ideas, or suggest different ways to tell a story. Sometimes a picture even nudges a piece in a direction I had not expected. It becomes another language to convey the same story, another way to communicate. And now the cost of this age-old form of communication, especially for those of us without any artistic talent, is basically zero.

Midjourney has recently added a video feature. Now, with a short prompt and a click of a button, I can create my own animation. Consider the example below. Of course, Midjourney is not alone; competitors like Veo3 for video are launching new possibilities almost daily. Where might this take us? The frontier of creativity is shifting rapidly. Now, anyone with a good idea and a few lines of text can direct a film, conjuring up new worlds for us all to explore. If only there was more time…

Alexandria

From conjuring up new worlds to rediscovering the worlds of the past, the next AI tool in my digital toolkit is a tiny app named Alexandria, launched in July by Zohar Atkins and his team at Lightning Inspiration. Alexandria brings together more than four thousand classic books, each with a concise AI summary and an embedded AI tutor named Virgil to guide readers through the text or answer questions along the way.

You’ll find Aristotle’s Nicomachean Ethics and Plato’s Republic, but also more recent works such as Adam Smith’s Theory of Moral Sentiments (which I am now revisiting) and John Maynard Keynes’s The Economic Possibilities for our Grandchildren. Each comes with a clear summary that distils the main arguments and offers essential context. And best of all, it’s all free.

ChatGPT

There’s a free version of ChatGPT, the next digital tool on my list, but I use it so often that I’ve opted for the full package at $200 a month. That may sound excessive, but a full-time research assistant is at least three times more expensive, and none would be able or willing to tolerate my endless questions at all hours of the day or night.

What do I use it for? Frankly, it’s easier to list what I don’t use it for. I don’t, perhaps surprisingly, use it to write emails or social media posts. The internet is already awash with AI-generated content – ‘AI slob’ – and I have no desire to add to the digital pollution. An email, after all, is still a personal message; I’d rather those words come from a human. If you send me something that feels machine-made, don’t expect a reply. At the very least, show the courtesy of crafting a more interesting prompt.

Where I do use it, though, is at the core of my own workflow – not in the way some second-year students might when writing an essay, but as a sounding board and editor. I’ll take a paragraph I’ve written and ask for suggestions to make it clearer, more concise, or less repetitive. Sometimes it spots ‘favourites’ I overuse without realising; sometimes it nudges a tired sentence into something sharper. (Case in point: I wrote the first part of the previous sentence, and ChatGPT wrote the second.) It might seem as if I’m surrendering my ‘voice’, but, as my (human) editor has long reminded me, my ‘voice’ often just reflects my poor vocabulary or the need for speed. (I’m sorry my letter is so long, Mark Twain said, I did not have time to write a shorter one.)

Good non-fiction writing is about clarity, economy, and the occasional well-timed witty remark. In this sense, ChatGPT is an ideal editor: cleaning up my Afrikaansisms, making my prose more neutral and therefore more accessible to a wider audience. (Di, if you’re reading: don’t worry – ChatGPT is still no match for a good human editor. Just like me, it still can’t tell the difference between ‘especially’ and ‘particularly’.)

It is also remarkably useful for sorting through ideas. I’ll often ask for ‘orthogonal’ directions or endings for a text, just to see what emerges. It’s like having a panel of experts ready to give instant feedback. Sometimes I use it to check my own logic, to uncover holes in an argument, or to see how someone from a completely different field might approach the topic.

For example, after writing this piece, I asked ChatGPT 5 to critique it. Here’s the response. If some of it is confusing it’s because I’ve made a few tweaks given the feedback.

Critique (as an expert literary non-fiction editor)

Your piece has a warm, conversational style and a clear through-line built on the theme of “exploration” in the age of AI. The pacing is generally good, with each section introducing a new tool and illustrating its role in your creative and intellectual work. The strongest moments come when you connect the technology to a broader historical or philosophical frame (e.g., “secondary orality”, the printing press parallel). Those moments elevate the essay from a product review to reflective literary non-fiction.

However, the piece could benefit from tighter integration between sections. Right now, each tool functions as a self-contained vignette, but the transitions can feel abrupt, especially between Midjourney and Alexandria, or Suno and Granola. The historical “exploration” motif that opens the piece re-emerges only briefly at the end; weaving it lightly through the middle sections would give the essay more cohesion. There is also some redundancy in your phrasing (e.g., “I love it… I have always craved” and “It is easy to imagine…” appear in similar emotional registers) which could be pared down for impact. Finally, a few anecdotes could be sharpened or linked more explicitly to your central argument about how these tools are changing how we think, not just what we can make.

Of course, ChatGPT is good for many other things too. It produces code that lifts my productivity tenfold. It is an excellent translator from Afrikaans to English. (I still prefer a human translator in the other direction, partly because Afrikaans is rich in metaphor and layered meaning, but that might change soon.) I use its ‘deep research’ feature in my research to discover papers and sources I might never have found on my own, especially in unfamiliar fields – or ideas, theories and approaches I would not have considered. And then there are the reports no one reads: the bureaucratic paperwork that fills academic life. If a report exists only to satisfy some administrative requirement, I’m quite content for a machine to write it for me.

ChatGPT is not perfect, not even the most recent version 5. But, in truth, it is already difficult to imagine work without it. It has quickly become an indispensable tool for me, one that will only get better.

And it is not the only one. If ChatGPT is the friendly editor at your shoulder, the next tool is the tough examiner who checks every equation and cross-reference. It is the only app in this list I hesitated to share, because it adds an almost unfair advantage to any researcher…