As I mentioned in previous posts, you do not need to be an expert in AI to embed it in your insurance company. But, you must have access to that expertise. And the closer it is to the CEO, the better. Call it Chief AI Officer, Chief Analytics Officer, etc. However, the position is not about mathematical or coding expertise, but to manage expectations, team leadership, project execution, and strategic input.

Avoid mixing data protection and data science in the same role. The skill requirements are different and it will not work. In addition, most candidates will be interested in only one of those topics so, if the same person looks after the two, one will be underdeveloped.

A common mistake is to require great technical and coding skills from the top AI job. There should be a team for that. That person’s job is strategy and management. It is the team that needs to be good with Python, machine learning, etc. Do not expect your Chief AI Officer to know the most about all aspects of AI in your company, that is not their contribution.

While sound mathematical and coding skills are important, the world of AI moves fast, prioritising intellectual curiosity in your team over coding mastery is a good long term approach.

Beware of skill shortcuts. For example, most data scientists can help with data engineering but it might not be their strength, or what makes them happy. You get better results if using the right skilled people for each challenge. It will make a difference in the long run. In addition, given the high demand for data scientists, they will not last long with you doing work they do not like.

Developing a career path for your data science team will help too. What career options do they have in your company? It will show that you take them seriously, and it will provide an understanding of what skills they have to develop to grow with you. Potential hires would find it interesting too.

Set up a budget for your team’s ongoing education, make sure they spend time reading papers, going to talks, and mingling with peers. That is how they get better at what they do.

As a non-technical leader, you need a team that you trust, and to understand what is hard, what takes long, what long actually means (weeks, months, years?), what happens if something goes wrong and its implications, etc. Also bear in mind that all AI and machine learning projects have a highly experimental component. There is no predefined solution to your problem so there will be several attempts until you get it right. Because of this, estimating how long a project will take can be very challenging.