AI strategy

The format is a “World café” bringing together business leaders from various sectors.

Selected questions

Of the seven questions identified by participants, two received the most votes and formed the core of the discussion:

1. How can we manage the rapid evolution of the AI ​​ecosystem, its offerings, and its practices? Only an iterative approach seems capable of achieving this.

2. How can we manage the impacts of such a transformation on people, on finance, and on the planet? Only a long-term approach seems capable of achieving this.

The two approaches are considered compatible.

Question 1 : speed

The key topics specific to the first question (iterative approach) are:

• The right balance between a bottom-up and top-down approach, made possible by a governance structure based on "the right meeting point" and delegation of decision-making to middle management.

• The unambiguous adoption of an iterative approach such as agile/sprint.

• A non-negotiable time-boxing framework.

• Communities of practice that enable the rapid flow of information.

Question 2 : impact

The topics specific to the second question (long term) are:

- Sharing a collective vision both within and outside the company. Participants believe this is primarily a societal issue, where politics, geopolitics, and the stances of key leaders are prerequisites for any sustainable action.

- Compensating for the absence of such a shared vision at the national or even humanity level (participants did not hesitate to use this term). Company leadership must define its own vision, explain it, share it, and engage its organization by providing input on all key issues: risk management, training, preferred levers, and visualization of an ideal operating model.

- Benchmarking. This is considered beneficial in this context: with peers and with other countries (North America, China, Africa).

- Combating the loss of intent and value in human decision-making. This must be orchestrated and supported by clear practical recommendations.

- Measuring the true cost of AI.
- Protection of personal and professional data.

- Data ownership management.

- Security.

Two examples to follow were mentioned: Michelin and Renault.