The differences in approaches to artificial intelligence cannot be explained by the level of equipment, access to talent, or even the quality of data, but rather by organizational attitudes that are found across businesses, regardless of sector or size.
However, these attitudes are not always expressed as such. They manifest themselves in the way AI is received and accepted, and explain why some initiatives take root while others do not, and why the same debates recur almost systematically.
The aim here is not to rank businesses, let alone say which ones are successful, but to describe archetypes as observed in the field.
In short:
- The differences in AI adoption between businesses are mainly due to organizational attitudes, regardless of the technical or human resources available.
- There are five archetypes of businesses when it comes to AI: opportunistic, cautious, fascinated, structuring, and defensive, each reflecting a specific attitude towards the use and integration of AI.
- These attitudes influence the way AI is experienced, supervised, or restricted, with concrete effects on the sustainability and consistency of the initiatives implemented.
- A single business may combine several stances depending on its business lines or circumstances, and it is often the contradictions between the actual stance and the stated discourse that compromise AI projects.
- Rather than aiming directly for an “AI First” strategy, it is vital to understand the business’s current stance on AI in order to build a coherent and realistic trajectory.
The opportunistic business
This is undoubtedly the most common archetype. AI is approached as one opportunity among many, based on locally identified use cases. Businesses experiment, test, and adjust. Initiatives emerge where a team sees an immediate benefit, often on an individual or small group scale.
This approach produces visible but scattered results. Some uses persist, others disappear because there is no explicit desire for overall consistency or attempt to scale up, and AI is accepted as long as it remains useful and non-engaged (What works today with AI without any particular effort).
But when it requires coordination or clarification of responsibilities, the momentum slows down (The problems that arise when AI scales up).
This is not an ineffective approach, far from it. It allows for fast wins without challenging the status quo, but on the other hand, it also creates an accumulation of heterogeneous systems that are difficult to understand and rarely provide structure.
The cautious business
In this archetype, AI is primarily perceived as a risk. Legal risk, reputational risk, operational risk, and so on. Initiatives are therefore closely supervised and often slowed down by successive validation processes.
This caution has its merits, as it prevents excesses and limits poorly controlled uses, reducing the effects of fads, but it also has a cost. Projects struggle to move beyond the controlled experimentation stage, and teams become weary.
In these organizations, what works is rarely what has been officially authorized. The uses that stick are often those that fly under the radar, far from formal governance mechanisms, with all the risks associated with shadow AI.
The fascinated business
Here, AI is imbued with strong symbolic significance. It embodies modernity and performance and is therefore preceded and accompanied by ambitious discourse and carefully crafted demonstrations.
But this fascination is often accompanied by a disconnect from reality. Applications are conceived as permanent proof of concepts and rarely as long-term solutions. We show, we communicate, but we industrialize little.
In this context, AI becomes a subject in itself. It attracts attention, but struggles to become part of operations. Worse still, when it begins to produce concrete effects, it comes up against an organization that has not been prepared to deal with the consequences.
The structuring business
This archetype is rarer, but it does exist. AI is seen neither opportunistically nor as a fascination, but as a capability to be gradually integrated into the functioning of the business. The uses are not all ambitious, but they are designed to last.
This translates into a focus on what enables the systems to survive over time. Monitoring of uses, continuous adjustments, governance that evolves based on lessons learned—nothing spectacular, but a form of consistency.
These businesses do not necessarily talk about AI First because they often do not need to. Their approach does not guarantee success, but it greatly reduces the likelihood of the most common failures.
The defensive business
Finally, there are organizations for which AI is seen primarily as a destabilizing factor. It challenges established expertise, internal balances, and sometimes positions of power, and the response is to limit its impact.
Initiatives are allowed as long as they remain peripheral. Uses that are beginning to produce tangible effects are slowed down and there is a focus on less sensitive objectives. AI is not rejected, but is contained within a “safe” perimeter.
This stance is not always conscious, and in fact is rarely so, but is expressed through a series of micro-decisions which, when taken together, prevent any lasting impact as soon as AI touches on subjects considered strategic.
Coexisting archetypes
These archetypes are not mutually exclusive, of course, and the same business will often embody several of them, depending on the role, profession, or situation. What matters is not the label, but the consistency between the stated position and the practices that are actually tolerated.
It is often these discrepancies that cause initiatives to fail. Not because AI doesn’t work, but because it is approached with contradictory intentions. We expect it to transform, while treating it as a marginal tool; we celebrate it, while neutralizing it as soon as it becomes engaged.
These archetypes do not tell us what the business should do, but simply allow us to see more clearly what it is already doing in order to respond appropriately.
Bottom line
These archetypes do not tell us what businesses should do in response to AI, but rather what they are already doing, often without clearly articulating it or even being aware of it. They reveal attitudes that are expressed less in discourse than in everyday decisions.
Above all, they show that trajectories in relation to AI are rarely the result of an explicitly defined strategy, but rather a tangle of local choices. AI does not impose a direction, but rather reveals how the business deals with disruptions to its existing balance.
At this stage, talking about becoming AI First without taking these positions into account amounts to ignoring reality in favor of a sometimes implicit intention. Before even asking how to become AI First, it is necessary to understand what kind of business one already is in relation to AI, and how far this position allows one to go.
It is this gap between stated ambition and stance that makes most discourse on maturity so misleading.
To answer your questions…
These differences do not stem from technology, talent, or data, but rather from organizational attitudes. These attitudes determine how AI is accepted, regulated, or restricted on a daily basis. Implicit trade-offs, rather than stated strategies, explain why some initiatives become established in the long term and why others fail despite comparable resources.
Opportunistic businesses experiment with AI as local opportunities arise, without an overall vision. This approach allows for fast and visible gains as long as the applications remain simple. However, the lack of coordination and scaling leads to scattered initiatives that are difficult to maintain and rarely provide long-term structure.
Excessive caution limits risks but significantly slows down projects. Initiatives struggle to move beyond the experimental stage, which discourages teams. Truly effective practices sometimes end up developing outside official frameworks, creating a disconnect between formal governance and actual practices.
Fascination often transforms AI into a topic of discussion and demonstration rather than an operational tool. Projects remain permanent proof-of-concepts, with little industrialization. When AI begins to produce concrete results, the organization is not ready to manage the consequences, which hinders its long-term implementation.
L’entreprise structurante intègre l’IA progressivement, avec des usages pensés pour durer. Elle privilégie la cohérence, le suivi des usages et une gouvernance évolutive. Cette posture ne garantit pas le succès, mais elle réduit fortement les échecs liés aux contradictions entre ambitions affichées et pratiques réelles.
Image credit: Image generated by artificial intelligence via ChatGPT (OpenAI)
Related articles:
AI First: lots of debate and few answers
Adoption and impact of AI: lessons (and limitations) from the latest McKinsey and BCG studies
Who is handling your artificial intelligence projects? Probably not the right people.
A business can be born “AI first”, but can it become one?
What works today with AI without any particular effort
Why AI initiatives always fail for the same reasons
