Responsible Handover of AI: Asking the right questions to not cause harm

Sense about Science will explain how they are supporting AI developers, research funders, scientists, and others involved in AI tool development. Their aim is to ensure tools are funded and used safely and effectively by following the Responsible Handover Framework. AI and data science tools are emerging faster than quality assurance and implementation knowledge. As a result, we risk stifling their scientific capabilities. In some cases, AI tools are being implemented inappropriately and inadvertently causing harm. In others, opportunities for positive impacts are missed when tools are omitted due to concerns about assurances and whether the right questions have been asked. As a tool is developed, there are key knowledge transfer or handover points where vital information about its likely strengths and weaknesses in different contexts should be discussed. But these conversations are prone to slip through the cracks. This is likely to occur when the number of users of a tool grows or there is a big difference in technical expertise between tool adopters and tool proposer. This results in a gap between high-level principles and technical domain specific guidance. This can be overcome with a general, practical governance framework. Join Sense about Science as they describe how the Responsible Handover Framework was developed, how the framework covers the risk governance gap, and how to formulate your own Responsible Handover strategy.