citiesabc, first_page
AI Overestimates Work Ethic in Major UK Cities, New Study Finds
12 Oct 2026
Research from Go Vista AI finds that ChatGPT, Gemini and Claude often rank high-profile university and business cities as harder-working than official labour data suggests, while smaller towns and districts are frequently underestimated.
5 October 2026 — Artificial intelligence models may be relying too heavily on reputation when asked which parts of the UK work the hardest, according to new research from Go Vista AI.
The study compared how ChatGPT, Gemini and Claude ranked British areas for work ethic against labour-market data from the Office for National Statistics, including average weekly paid hours, the share of people working 49 hours or more, employment rates, full-time employment and economic activity.
The biggest gaps appeared in well-known cities and university centres. Exeter, Reading, Birmingham, Manchester and Oxford were among the places AI ranked substantially higher than the underlying labour data.
Exeter is the most overrated by AI
Exeter recorded the largest gap in the study.
Across the three AI models, the city ranked 5th for work ethic, but the real-world data placed it 41st, a difference of 36 places.
Workers in Exeter average around 32 paid hours per week, while 8.8% work 49 hours or more. Its employment rate stands at 74.8%, with a full-time employment rate of 67.8% and an economic activity rate of 77%.
Go Vista AI argues that the city’s strong university presence and technology reputation may be influencing how language models interpret its productivity.
Reading’s reputation outpaces the labour data
Reading ranked 4th in AI estimates but only 38th in the underlying labour data, making it the second-most overrated area.
The study reports an employment rate of 72.7%, while only 7% of workers put in 49 hours or more.
Go Vista AI suggests Reading’s association with the M4 technology corridor and proximity to London may contribute to a “high-achiever” image that AI models reproduce, even when the local workforce statistics are less exceptional.
Birmingham and Manchester also rank too highly
Birmingham was the most overrated major city in the study.
AI models placed it 18th overall, while official labour data ranked it 43rd. The report says Birmingham had the lowest employment rate in the dataset, at 67%.
Manchester showed a similar pattern. The models placed it around 20th, but the labour data ranked it 44th. Average weekly paid hours in Manchester were around 31 hours, while the employment rate was 74.9%.
The findings suggest that large-city reputation, industrial history and cultural visibility may influence AI-generated rankings more than less visible labour-market metrics.
University cities appear to be consistently overrated
The research identified a wider pattern around university centres.
Oxford ranked 16 places higher in AI responses than in the real-world data, while Exeter showed the largest overall gap. Cambridge was also cited by Go Vista AI as a city that AI tends to rank more favourably than the labour data would suggest.
That raises an important question about how generative AI systems form judgements about cities.
When asked broad questions such as “where do people work hardest?”, language models may draw on common associations with prestige, education, industry, technology and economic importance rather than direct statistical evidence.
In other words, they can reproduce reputation-based narratives rather than labour-market reality.
Smaller towns are often underestimated
The reverse also happens.
Go Vista AI found that Basildon, Winchester, King’s Lynn and West Norfolk, Crawley and Newport were among the areas AI underestimated most heavily.
Basildon had the largest positive gap, ranking 40 places higher in the real-world data than in AI responses.
According to the study, workers in Basildon average 34 hours per week, while more than 11% work 49 hours or more.
Crawley was another example. The area recorded an employment rate of 90%, the highest in the study, but AI ranked it only 27th.
A Go Vista AI automation expert said:
The most underrated areas tend to be smaller towns and districts that don't carry the same name recognition as cities like London or Manchester.
What the study says about AI bias
The research is not evidence that the models are “wrong” in every sense when discussing work ethic. The study itself defines work ethic through a specific combination of labour indicators.
But it does show that AI-generated reputational judgements can diverge sharply from structured data.
That matters because generative AI is increasingly being used for research, planning and decision support across businesses and public organisations.
If a model is asked to rank cities, regions or communities without being given a reliable dataset, it may lean on patterns in its training data, including media coverage, stereotypes and familiar narratives.
For urban policy and local economic analysis, that can distort perceptions of places that receive less national attention.
The danger of mistaking visibility for productivity
The strongest theme in Go Vista AI’s findings is the difference between visibility and measured labour performance.
Large cities such as London, Birmingham and Manchester dominate public discourse because of their economic importance, cultural influence and media presence.
University centres such as Oxford, Cambridge and Exeter benefit from reputations built around education, research and innovation.
Smaller towns and districts often receive far less attention.
AI systems trained on vast amounts of public text may therefore absorb those imbalances and reproduce them when asked subjective questions.
The result is a kind of reputational bias: the places people talk about most can appear stronger than the data supports, while less visible areas are overlooked.
A useful reminder for cities using AI
For councils, employers and local development organisations, the research offers a practical warning.
AI-generated rankings should not be treated as substitutes for official statistics.
Models can be useful for summarising data, identifying patterns and exploring scenarios, but the quality of the answer still depends on the evidence supplied to the system.
When questions involve employment, productivity or economic performance, official datasets remain essential.
The study therefore highlights a broader point for cities adopting AI: the technology may be powerful, but it can still inherit the assumptions and visibility biases embedded in the information it was trained on.
About Go Vista AI
Go Vista AI is a digital marketing and automation agency that publishes research on artificial intelligence, search behaviour and digital trends.
Share
Sara Srifi
Sara is a Software Engineering and Business student with a passion for astronomy, cultural studies, and human-centered storytelling. She explores the quiet intersections between science, identity, and imagination, reflecting on how space, art, and society shape the way we understand ourselves and the world around us. Her writing draws on curiosity and lived experience to bridge disciplines and spark dialogue across cultures.
