AI exploration space

• 6-16 people
• 1 day, or two half-days
• no technical background required
• developing AI ideas with cards

In most organisations, AI is discussed at two separate tables: at one sit the people who could do it technically, at the other the people who know what it is actually about. The AI exploration space puts both at the same table, and gives everyone the same tool.

It builds on the AI Toolkit: a set of cards describing the building blocks of machine learning in a way that makes them combinable without jargon. Data, methods, output forms, risks, each card is an element you can put on the table, move around and take away again. Combining them produces concrete application ideas, step by step.

1. Arriving: what connects those present to the topic, and what do they fear?
2. Getting to know the building blocks: the card set, explained by example.
3. First combination: small groups produce three to five ideas.
4. Hardening: each idea is tested against data, law and benefit. Some drop out, and that is a result.
5. Storyboard: the most viable idea is visualised, from trigger to effect.
6. Situating: what would the next step be, and who would have to take it?

Shared understanding and a robust concept prototype, usually a storyboard. That serves as the basis for a pitch, a project proposal or a first technical specification.

Explicitly not finished code or an architecture. And possibly the finding that AI does not help here. In that case you have invested one day instead of a project budget, which is the cheapest conceivable outcome.

A mixed group: department, IT, management, ideally also data protection and staff representation. Homogeneous groups produce homogeneous ideas.

Your own data is helpful but not required. The day works just as well with example material; with your own holdings it gets more concrete. Nobody needs a technical background: only as much is explained as is needed to take part in shaping things.

Anyone looking for a systematic introduction. The workshop Understanding Machine Learning is meant for that, going through the field in order rather than by example.

And projects where it is already settled what will be built. An exploration space sorts possibilities, it does not confirm a decision.