As artificial intelligence (AI) is being used more and more by transport firms, Fleetcheck has warned against the use of popular general use tools which can produce wildly unreliable outputs.
The fleet software firm said generative AI tools such as Claude and Chat GPT can be extremely effective but also frequently incorrect, with wildly unreliable outputs ranging from minor inaccuracies to completely misleading information.
Callum Haymon-Collins, chief operating officer, said:
“We’re seeing fleet managers across our user base start to use AI extensively and getting a good feel for the reliability of its results. There is no question that, sometimes, its responses are excellent and rapidly create insights from data that would be difficult to achieve through any other means.”
But he also warned of occasions when its outputs lie somewhere on a line between ‘misleading and dead wrong’.
“For example, I recently uploaded some fuel card data into a popular AI tool and its key finding was that I should closely question the provider about pump price rises between February and March, strongly implying there could be some commercial malpractice underway. It didn’t spot that the war in Iran had caused the issue, which is obviously a fundamental error.”
And problems of this type are likely to become more prevalent over time, as the number of hallucinations produced by generative AI increases, said Haymon-Collins.
“We’re in an odd situation where the AI models are getting better but the data they use is getting worse because the information on which they train is now largely based on AI output.
“OpenAI’s own publicly published research shows its latest reasoning model now creates hallucinations 48% of the time compared to just 16% in older versions. This is not a bug that can be solved but a feature of the technology.”
