What’s been your biggest challenge in implementing practical AI solutions was it data quality, stakeholder buy-in, integration with legacy systems, or something else entirely?

For us broadly in the UK public sector-  data quality, unstructured data and data silos are the biggest challenges.

One of our biggest challenges is proper scoping. It's easy to say "use AI" but defining what that actually means and implementing appropriately can be difficult. It's important to fully understand the use case before sourcing or building a solution. AI can be a great tool, but it's not always the best tool for the job.