For what its worth, some thoughts on AI from someone who did a computing degree followed by a humanities masters …

Summary: after the noise and clickbait has faded, the implications for creative content makers will be ‘a flight to quality’ where brands or names will earn value by their reputations – hard won and easily lost.

My laypersons sense of AI in terms of creative content is that AI will become like ‘the channel five of the internet’ – low common denominator, time-filling, bland, forgettable. Cheap brands will use AI to make buckets of unremarkable stuff. Ask ChatGPT to write a treatment for a programme on Benny Hill – it won’t exactly inspire the awards panel.

Where AI comes into its own will be in so-called narrow applications: better speech recognition, better scans of medical images, better organising of large amounts of raw data without having to program a database. It’s been around longer than the public realise, for example in the handwriting recognition of historic documents for USA family history websites, still commercially sensitive.

And I think it will be the “new narrow” applications in the next few years that will surprise us most – there are already researchers working on AI for court hearings and decisions with the promise of no discrimination in sentencing; and on medical triage on phones before you see a GP, or in many countries, an advanced nurse with prescribing powers. Imagine uploading every scrap of paper from the Stephen Lawrence police files, and the AI summarising the weaknesses and gaps, and then for other cases in real time directing the police investigation. “AI SIO” to coin a phrase. Not impossible, AI is already passing the New York Bar Exam with consistent 80+% marks.

Or imagine a professional video editing suite where one written instruction can “remove the traffic noise from the left audio track, leave it in the right” or “add falling rain to the window pane in scene 7”.

But for now, if you just ask the internet how many types of AI are there, you will discover that there are 3, 4, 5, 7 or 9. (If it wasn’t for the ‘4’ and ‘9’ I would suspect a pattern based on prime numbers.)

Final policy point: the degree to which quality brands can protect their content from adulteration may determine their prospects. This isn’t quite the same as protecting their content from being scraped into an AI grinder. Like a hologram on a banknote to protect against forgery, it will be the digital hologram on quality content that protects value. For my money this will be done using blockchain tech as a protector of the provenance of premium content. Even of digital currencies are just a 2020s bubble-scam the underlying tech is sound.

The interesting question is which brand will become the guarantor of provenance – the creative Bank of England equivalent which will sit behind all the retail brands? As a publisher I buy unique ISBNs for ‘my’ books by the dozen, I register the details with an international database as each book emerges into the world, and Google Books etc all use this database to find fraud.

TB