Research highlights limitations of AI in workplace consulting, introducing 'trendslop' concept
Economists Mariana Mazzucato and Rosie Collington criticize the consulting industry for providing misleading guidance, suggesting that AI may not be a viable alternative. A study from Esade Business School reveals that AI models often rely on popular jargon rather than tailored advice, leading to a phenomenon termed 'trendslop.'
Economists Mariana Mazzucato and Rosie Collington argue that consultants often provide dubious guidance and can worsen dysfunction in both government and private sectors.
Their book, 'The Big Con,' claims that consultants emerged in a post-Reagan era of reduced regulations, creating an illusion of value while draining resources.
Research from Esade Business School indicates that large language models (LLMs) like GPT-5 and others tend to favor buzzwords over contextually relevant advice, leading to the term 'trendslop.'
In a study involving 15,000 simulations, researchers found that LLMs clustered their responses around similar strategies, lacking diversity in solutions.
The study warns that relying on AI for consulting may result in generic advice rather than tailored solutions, posing risks for business leaders.
LLMs exhibit biases based on their training data, often favoring certain phrases and concepts, which can lead to unproductive or harmful outcomes.
While the researchers acknowledge that LLMs can help generate alternative solutions, they emphasize that AI should not replace human judgment in decision-making processes.