Nicole JunkermannReading · Travel · Writing

Ideas

Nicole Junkermann on Why Smaller AI Models Suit British Firms

Not every question needs the biggest answer. Nicole Junkermann on why proportion may be the most useful idea in business AI.

By Nicole Junkermann · Published

Back to the blog
Nicole Junkermann, portrait, AI in British business

A good deal of talk about artificial intelligence assumes that bigger is always better. Nicole Junkermann prefers a more modest question, and perhaps a more British one: how much machine does this particular job actually need?

Nicole Junkermann on matching the AI model to the job. Transcript and notes

The size of the job

In her short film she says that not every task needs the strongest model, and that a simple question can take a smaller, faster route. It sounds obvious and is surprisingly easy to forget. Sorting the morning's enquiries, drafting a routine letter or pulling a date from a long document are modest jobs. A smaller system tuned for that kind of work could do them perfectly well, more quickly and with less computing power along the way. Nobody hires a removal lorry to bring home the shopping.

The shape of British business

Much of British commerce is made up of small and medium-sized firms: regional manufacturers, family distributors, professional practices, specialist makers whose knowledge has built up over decades. Such businesses rarely have a technology department of any size, and few would want one. What they tend to have is a clear set of recurring tasks and a healthy suspicion of anything expensive. A smaller model fits that temperament. It may be simpler to understand, cheaper to run and easier to keep close to a firm's own records, which matters to anyone who looks after a client's information with care.

Judgement, in proportion

There is also an argument from character. A tool that does one job within clear limits is easier to check than one that attempts everything, and it leaves the person using it firmly in charge. Good AI workflow, as she puts it, means matching the tool to the size of the job. The same instinct runs through her thinking on skills as the slowest investment and on what thoughtful AI could mean for British innovation. A firm that begins small may learn what it actually needs before committing to anything larger, and that knowledge could prove worth more than any particular piece of software.

A sensible place to begin

None of this says which technologies are going to prevail, and very large models may well keep doing remarkable things. The point is proportion. An owner wondering where to begin could do worse than list the dullest tasks of the week, choose one of them and ask what the least elaborate tool able to do it might be. Nicole Junkermann enjoys the plainness of that exercise, and it happens to suit a country that has often done well by being practical rather than grand.

November feels right for it, with the year's plans being quietly reviewed and next year's not yet fixed. The small improvement, properly made and properly looked after, has a long history in Britain. Smaller models could turn out to be a natural home for that habit, and that is a cheering prospect for the firms and the people who work in them.

This article is a personal reflection and general commentary. It is not financial, investment or professional advice.

More from Nicole Junkermann

For more from Nicole Junkermann, read why UK health tech rewards careful stewardship, who is accountable when AI supports a clinical decision and Britain's quiet strength in satellites and automation. The Notebook keeps the smaller thoughts.

← Back to the blog