AI: Empowering Human Judgment with Safety, Empathy, and Trust

Our Managing Director, Dominic, recently completed a Data & AI course at Manchester University. Here’s his reflection on the experience and the key insights he gained:

Last week I went back to University, specifically Manchester University, to complete a 4 day course on Data & AI. The course has strengthened both my technical understanding of AI and, critically, my awareness of how to use these tools safely within Timeout. A key message was that effective AI begins with well‑framed, context‑aware questions, ensuring the technology serves real organisational needs rather than driving unnecessary complexity.

Equally important was learning how AI models can fail. We explored issues such as hallucinations, where generative AI presents incorrect information with confidence, and bias, which can arise from incorrect or incomplete training data. These risks are especially relevant in a care environment, where fairness, accuracy and accountability directly affect how we view information.

Safe AI usage, therefore, requires robust governance. This includes validating the information that we receive from AI rather than accepting it at face value, keeping a human at the centre of the decision, and checking for drift, bias, or unexplained behaviour. We also covered adversarial attacks, data poisoning, privacy breaches, and prompt‑based manipulation, which are all reminders that technical safeguards don't guarantee 100% safety.

As I have said before, we should use AI to support us, not replace our own empathetic, human judgment. We need to ensure we understand the models we use and where the information comes from. We need to be clear through our policies and procedures that we are compliant with data protection and that our people understand what to use tools for and their limitations and never forget to check. Ultimately, with the work we do, it aligns with our safeguarding principles, ensuring that we utilise these tools with accuracy, dignity, and trust. 

Overall, the course has given me the confidence to identify beneficial uses, such as predictive maintenance scheduling, pattern detection, rota scheduling or improved reporting whilst ensuring guardrails are in place. The systems we all use every day will start to incorporate agents and chatbots there to assist us, but not replace us, also considering how children will interact with this new world of information representation. Safe AI adoption is not just a technical task but an organisational responsibility, and I am certain that as we step into this fast-moving era, it will allow us to be more human and, to that end, we do not lose our inquisitive nature to ensure we keep ourselves and those we care for and educate safe.