‘The establishment, verification and use of management systems are spot on for use of AI’

AI “might be the salvation to humankind, or the opposite, as some fear”, says Lars Riisberg, CEO of RINA Digital Solutions, in this Lloyd’s List podcast. Today, with large language models such as ChatGPT giving the general public some understanding of AI’s capabilities, he says that AI “will surely change the way we work dramatically”. During his career devoted to data and its application that goes back to his university thesis in 1989, which looked at using AI in the automation industry, Riisberg has developed a positive outlook about AI. He is confident that, for shipping, it will have a positive effect on many current objectives, including decarbonisation, compliance and safety. In his opinion, digitalisation and the use of AI will help shipping’s decarbonisation agenda and general operational efficiency. In particular, “the establishment, verification and use of management systems are spot on for use of AI,” he says. Yet all these benefits rely on data and one of his podcast themes revolves around how AI itself can be used to ensure that data is gathered in reliable ways. He listed some key data sources and acknowledged that “they all have their challenges when it comes to quality.” In his podcast, Riisberg mentions some practical instances where he believes AI can make a positive change, for example by using it to guide crew in how they carry out and report tasks. At an operational scale, he said that AI is already being used to establish reference performance standards in different sea states and to highlight deviations in fuel consumption compared to those standards and alert the crew. He also provided examples of where AI can be applied to management tasks but said that RINA's experience has given some insights into the pitfalls of AI. “AI is generally very ‘polite’ and will try to come up with an answer… even when there is no good answer to a question,” he said, so “you will still have to be able to evaluate the answers and take your own personal judgement of the correctness of the answers.” Looking ahead, he predicted that it will become possible to interrogate data and written material by speaking to it through an AI interface and concluded “being able to ask questions about technically structured data… will be a new focus area for AI in the coming years.”