In the city in which I live, Leeds, in northern England, there is almost constant construction work going on: redesigning traffic junctions, constructing cycle lanes, building pedestrian bridges with gradual inclines, improving access to the main train station, creating pedestrianised city centre areas, planting trees and creating small parks. It’s progress, and I welcome it, but it is important to note that this good progress is correcting bad ‘progress’ of several decades ago.

In the 1970s, Leeds was dubbed ‘Motorway City’, and everything was designed for the convenience of car drivers. Blocks of terraced houses and historic buildings were cleared to create dual carriageways, flyovers and huge junctions. Suburbs became cut off from other suburbs, communities became fractured, air quality declined, green spaces were almost non-existent, noise pollution increased. It became easier to get to the city, but few people wanted to.

There was scant regard for pedestrians, and none whatsoever for cyclists or wildlife. The plans were driven by the latest technology of the day, reflecting the rise in personal car ownership in the 1950s and 1960s. It was a historic error of planning and policy. Only after many years of expensive correction can the city now attract visitors, as well as becoming a pleasant place in which to live.

Are we repeating the mistake with the AI revolution? The very last people who ought to be in charge of the roll-out of any technology are the technologists themselves.

They have a tendency to promote what the technology is capable of, rather than what it should be for – which is not the same thing at all. Ironically this often results in sub-optimal use of the potential of the technology – the subject of which dominated this year’s London Book Fair, as my colleague Susan Feldstein wrote recently.

We are told that AI will do this. AI will change everything. AI is super-powerful. This is only half right. AI will only achieve the claims made for it with intelligent human direction. Also, online platforms are now exceptionally prone to corruption by toxic AI bots, and dealing with these now takes up a growing amount of time and resources of legitimate companies, to try to repair customer trust.

Technology is best when used in harmony with humans. Google DeepMind has found that the accuracy of using large language models (LLMs) rises from 16% to 99% when expertly guided by humans equipped with what are now known as fusion skills. Despite this hard evidence on the value of human intelligence, there is a looming legislative threat that LLM manufacturers will be allowed to continue to hoover up creative content, in an exemption from copyright law, to train their models and churn out derivative content. This, we are told, is ‘progress’.

Technologists will always argue that it is the nature of technology-driven capitalism to be disruptive.

There are two counter-points: technological disruption is not always positive, and not all positive disruption is technological. In my lifetime, there have been two major disruptors that have changed the way in which we live: computers (as a generic term, from mainframes to the smartphone and AI), and a four-piece musical group founded in Liverpool more than 60 years ago known as The Beatles.

The Beatles gave a social licence to people without a privileged background to set up their own band, or other creative outlet, to teach themselves – not only white working class British boys with guitars, but much more widely: Whoopi Goldberg cites them as an inspirational influence, for example.

They encouraged equality and acceptance, through a refusal to accept inferiority of certain social classes or accept racial discrimination. The Beatles refused to play US theatres if they segregated the audience – this was written into their contracts – so they accelerated the progress of civil rights in the US and beyond. They championed Indian music and culture, at a time of extreme hostility from many other British white people to immigrants from south Asia.

Above all, they created the soundtrack for a generation and beyond, and inspired thousands of other songwriters and other creative people along the way, enriching the daily experience of everyone.

The Beatles loved technology, but they made sure that they, the creative artists, remained in control. Since the introduction of drum machines and autotune in the 1980s and 1990s, there has been a deadening effect on the quality of recorded music, with a dull uniformity creeping into the sounds of different artists.

There is now a reasonable fear that the tech bros want to cover many of our social and creative spaces with digital concrete, as more AI-written ‘novels’ enter the market. There is a better way: ensure that AI remains a tool, and place humans at the heart of all progress.

At UG, we embrace technology. AI records our interviews and transcribes notes for us. It can be used for revision of manuscripts. At least with an AI review, you know the LLM has read the whole text, which doesn’t happen with the lazier human reviewers. But it’s stronger on structure than style, and by definition it’s derivative, as it’s limited by its dataset.

Like the Beatles and the Google DeepMind research department, we seek to keep the humans in control, to maximize the creative potential. Humans are capable of judgement and emotional intelligence, we can draw upon the lived experience, and we can be genuinely original and creative. The dataset of an LLM is a limitation as well as a resource. Making the dataset bigger doesn’t address its inherent weaknesses.

There’s a similarity with the technology for city transport: roads should take us to human communities, not replace them.