AI driving experts from online community – study

When AI can instantly generate expert-looking answers, people who have spent years developing specialist knowledge may feel their contributions are no longer valued or recognised, says Business School researcher Dr Kenny Ching.

Analysing activity on Stack Overflow, the world's largest online community for software developers, Ching found respected, high-reputation users started withdrawing from the platform at accelerated rates in 2022, when generative AI tools became more widely available.

“As AI-generated content becomes more common, people might feel their expertise and effort no longer stand out or are valued. Some stop contributing altogether.”

Worryingly, says Ching, the people most likely to withdraw are often knowledgeable and capable contributors.

“They aren't leaving because they can't compete with the technology; they’re leaving because their hard-earned expertise is no longer distinct from a chatbot’s answer.”

Ching says although his research focuses on Stack Overflow, similar scenarios are likely playing out in other areas.

“This isn't just about coding platforms. I argue this exact same dynamic is discouraging genuine effort in classrooms, corporate workplaces, and scientific communities. The long-term risk is that by crushing the incentive to show genuine effort, AI might actually truncate the formation of future human expertise.”

Created in 2008, Stack Overflow is a question-and-answer site mainly for computer programmers. People can vote questions and answers up or down, similar to Reddit.

Users of the website, which has around 23 million registered accounts, can earn reputation points for well-informed answers and valued contributions.

For example, a person with 10,000 reputation points has provided answers that the developer community repeatedly judged to be accurate, helpful, and well-explained.

Ching analysed 24,304 contributors over 17 months.

He found that following 2022, which saw the widespread availability of generative AI tools, including the launch of ChatGPT, the rate at which higher-reputation users left the platform sped up over time – steadily closing the gap with less established contributors, who tended to leave first, but whose departures didn’t speed up in the same way.

AI, says Ching, can make expert and non-expert contributions look increasingly similar, reducing the visibility and value of genuine expertise.

He calls this ‘signal compression’.

“On Stack Overflow, the withdrawal of genuine contributors is real, measurable, and accelerating, and the most affected are often those whose authentic expertise is most valuable and hardest to replace.

“If everybody can create a good quality response or output using AI, some people may think, 'Why should I make an effort to share my expertise and participate?’”

His study argues that people with ‘high ability,’ who have typically invested most in their knowledge, face the sharpest devaluation of that investment as a result of AI, and they eventually stop participating and leave the platform.

Read Dr Kenny Ching’s working paper, When Effort Stops Signalling: AI, Signal Compression, and the Withdrawal of Authentic Performance.