Are we there yet?: keeping the human in the loop

Theme: Responsible, sustainable working | Making impact | Dealing with change | AI and beyond

Government projects integrating AI decision making into their user journeys have taken off.

But have we done enough work to understand what AI automation really looks like in high risk government spaces and if, when and where we should keep humans in the loop?

We need to do more with less, now more than ever. In the context of lower UK economic output, but the higher demand for public services (with larger populations) we need to make sure we use AI to get our decision making right, keep costs down, reduce mistakes and stop the potential public scandals caused by AI inaccuracies/a lack of human oversight.

Laura will share learnings from working in an exclusively AI-orientated team in government, playing back the practical challenges of integrating AI decision making in previously human-led spaces (internal to the organisation) and what potholes we need to avoid. 

In particular, she will share a model for understanding risk when deciding on removing the human from the loop when involving-AI generated decision making for service workflows. 

The session will include recommendations on appropriate uses and considerations for the audience to take away and apply in their own work places.

Participant takeaways:

• How to consider business needs and requirements when dialing up AI automation and dialing down human involvement
• A tool to help to understand how much a human needs to be kept in the loop, and via what methods
• Knowledge of how we can make sure we do not have skill attrition or dilution from key internal users when integrating AI in expert 'judgement call' spaces
• How 'mindful friction' helps bring our users (internal or external) into the loop in a way that brings the best out of their decision making and improves their judgement of algorithmic outputs
• Ways to understand or anticipate 'inevitable' human responses to AI algorithmic outputs