The Great Annual Biography Update

It’s that time of the year, when artist managers around the world edit the promotional biographies of their protégés. Can AI help?

August is traditionally the month when classical musicians get to have fun with their buddies at music festivals in beautiful places, but spare a thought for their managers (and the people, like me, who help them). For this is the time of the Great Annual Biography Update.

I wrote on the subject of musicians’ biographies in 2020 and my advice is largely the same. The challenge of the old-fashioned long biography – the one that goes on managers’ pages and in concert programmes – is that these texts face a variety of audiences, each with different needs. When you write or edit them, you have to triangulate between these. The audience likes a good story, the manager and artist need to impress, and the promoter just wants to bung something into a template quickly, alongside other biographies.

It’s hard to satisfy everyone. The audience might want to hear funny or moving stories about an artist, or how much they love their cat, or at least some detail that distinguishes them. The manager wants to list the major orchestras they’ve played with, in a perfect balance of countries and formats, as well as the big-name conductors they’ve worked with, according to some cryptic manager–promoter code. The marketing executive just needs 700 words of clean text that reads well. What you read in a programme is inevitably a compromise.

What has changed since I wrote in 2020 is that we now have AI to help us. I’ve written about my fascination for finding good uses for AI and I’ve come to believe that writing biographies is an ideal task for it. After all, its greatest weakness is how generic it sounds when generating text – and that’s exactly the right style for biographies. They shouldn’t sound too personal or quirky – save that for the artist’s website and social media. They should be informative, well structured, grammatically correct, accurate and clear. These all play to AI’s strengths.

On the other hand, the subtext and detail of a good biography should be highly specific. An artist and their manager should know their strengths and/or what makes them different from the next artist on the list. These things can be hard to articulate, but if you can even just blurt out three or four thoughts, AI can usually craft it into something that forms the basis of an opening paragraph that you can work with. Alternatively, you can ask it to search for the defining features of an artist and it will draw on reviews and programmes from the internet to create something that you might have thought of.

This is not a passive process. You can’t necessarily accept its first response. AI serves as a tool: imagine it as your personal subeditor, and not your ghostwriter. You can spar with and challenge it, trying new structures and emphases, or finessing the language and detail. And yes, this might take as long as doing it yourself, but the goal here is to have a better biography, not necessarily to save time.

In an ideal world, every concert listing and programme would have a beautifully written bespoke artist’s biography relating to the repertoire, venue, city etc. AI might also help with this eventually, but with the resources and time pressures facing the classical music world, this is unrealistic. It can at least help improve the long biographies that form the basic narrative of how an artist is known. It can also make life a little easier for people in this situation who dread writing – as well as those of us who dread writing badly.

A few points of advice from my experience:

Do

• Copy in your current biography and ask it to critique and improve it.

• Ask it to generate a new biography based on what it can find online.

• Be specific in your demands and feedback – the more detail you give it, the better the answer you get.

• Work on the first paragraph as the most specific expression of you as an artist, feeding in what you think are your essential qualities, experiences or musical tastes.

• Demand that it takes out all hyperbole, jargon and inelegant phrasing – it usually finds an improvement, although these words sometimes creep back in later.

• Ask it for sources for any information it finds online to avoid introducing errors (and always check them).

• Keep going until you’re happy.

• Come back to it a day or two later – sometimes the process can drive you a bit copy-blind.

• At the end of the process ask it to fact check and proofread it and even critique it all over again.

  
  

Don’t

• Don’t trust any information it pulls from anywhere – even your current biography. I’ve seen it invent orchestra names and mangle ideas.

• Don’t be taken in by its flattery. It might tell you something is wonderful and when you point out a flaw, it’ll backtrack speedily.

  
  

I asked ChatGPT for the top ten things it thinks I should include in these instructions. This is what it says (bearing in mind that it knows me well by now so this is filtered through the prism of my own opinions):

1. Don’t start from scratch. Give AI your own verified material; use it to shape and edit, not invent. 2. Create a source document. Include the facts, current projects, repertoire, collaborations and achievements you actually want to draw on. 3. Define the purpose. Tell it whether the text is for a manager’s website, concert programme, press release, or something else – and who will read it. 4. Give it something distinctive. Explain what makes this particular musician interesting rather than asking for a generic “engaging biography”. 5. Tell it what to leave out. Ask it to select the most revealing details rather than turning the CV into prose. 6. Ban the clichés. Tell it which words and phrases you dislike – and ask for precise, concrete language instead. 7. Work on the opening separately. Ask AI to identify the artist’s central idea and suggest several possible first sentences. 8. Ask it to be an editor, not just a writer. Have it identify repetition, generic claims and what a reader still wouldn’t understand about the artist. 9. Fact-check everything. Especially names, dates, engagements, awards, repertoire and anything AI may have filled in itself. 10. Keep the human voice and judgement. Use AI to organise, challenge and refine the material, but the musician or manager should decide what the biography actually says.

It also offered this useful insight (if only everyone were this self-aware!):

What AI is genuinely good at

→ restructuring, shortening, adapting tone/length, identifying repetition, generating alternatives, interrogating a draft

What AI is bad at

→ deciding what is artistically interesting, knowing what is true, understanding a musician’s personality, avoiding generic arts language

What the musician/manager needs to supply

→ facts, priorities, personality, positioning and editorial judgement