Manchester Statistical Society’s AI Study Group held its third meeting with a talk titled “AI, knowledge and behaviour” by Dr Richard Whittle, Research Fellow at the University of Salford. Dr Whittle delivered a very insightful and all-inclusive talk, which brought together many themes such as generative AI, the capabilities and use of generative AI in different contexts, dark patterns and the need for digital skills to take advantage of potential that AI offers.

Beginning with the essential features of generative AI tools such as Chat GPT, Dr Whittle took the audience on a tour of summarisation and response features. In this context, Dr Whittle demonstrated the importance of prompts and showed examples of prompts that produce better responses. For instance, interacting with a generative AI tool as if the tool were a human and using emotional cues tend to produce better results. Dr Whittle argued that this is largely due to the training that has gone into tools. Since the language models the tools are trained on use human generated content, they can spot patterns in responses given to different types of questions.

The interactiveness and ease of generative AI tools, not to mention the media hype, have made them very popular. We are, therefore, in the midst of an AI summer – a period characterised by developments and optimism around AI – but in the past decades, there have been both AI summers and AI winters. While some of the hype may wear off in the coming months and years, the impact generative AI has had in various occupations and settings is quite marked.

The advent of generative AI presents numerous challenges. Having the skills to engage with the tools is one of them. One way to enable individuals and businesses to take advantage of these developments is a programme of substantial digital upskilling. That, however, is expensive. It could also be disruptive if, for instance, a business had to take staff time away from core operation and put them through training. In this concluding remarks, Dr Whittle gave participants a lot of food for thought on what and how society might adopt and adapt generative AI tools. On the whole, it was an excellent exposition by on the current state of AI.