Business leaders are accustomed to engaging with people – on the board, among consultants or with colleagues – in meetings where meaning and value emerge from the interplay of different experiences and perspectives. Today, however, leaders are discovering a new interlocutor: generative artificial intelligence. Not as a collaborator to whom they can delegate the drafting of texts or reports, but as a cognitive partner, capable of opening up spaces for strategic reflection and helping to create new meaning for the organisation.

In general, the debate on generative AI focuses on productivity: greater speed, more automated tasks, or more lines of code. AI is essentially seen as a tool for performing operational tasks a little better and faster. Alongside this dimension, however, there is another, concerning its capacity to generate new strategic perspectives together with human beings.

Strategic opportunity, immature practice

Elisa Farri and Gabriele Rosani (2025) note that many managers have begun using AI as a strategic interlocutor. A partner with whom to assess perspectives, weigh up pros and cons and stimulate reasoning in support of leadership. Yet, although expectations are high, only a minority say they currently possess the skills needed to use LLMs in this way. To bridge this gap, the authors have published the HBR Guide to Generative AI for Managers, which offers practical guidance and prompts to make interaction more effective. In it, they also observe that, in most cases, the use of AI remains sporadic and superficial: managers often stop at the first level, delegating the initial formulation of their thinking to the algorithm, with the result that the cognitive level is lowered and ownership of the project diminished. How, then, can we move beyond this superficial use?

Thinking more, not thinking on our behalf

A creative, strategic use of interaction must be intentional: it is a process of deeper exploration, with a commitment to an outcome. A process in which the leader persists, provokes, reformulates and takes the conversation as high as possible. AI thus becomes capable of generating more developed, conceptual thoughts, moving ever further away from the merely practical.

In 1950, Alan Turing suggested that the point was not to ask whether a machine could think, but whether it could sustain a dialogue indistinguishable from a human one. We have arrived there. The paradox today is that the real question is not whether AI thinks, but whether it compels us to think more deeply than we would have done on our own.

Conversing with strategy

Academic research confirms that this use of AI is no longer a hypothesis but an established practice. David Rockmore (2025) has shown how many scholars already use ChatGPT in brainstorming, not to receive ready-made solutions but as a catalyst for introspection: AI forces them to look at their own thoughts from a different angle, making implicit assumptions and unexplored possibilities visible. He writes: This is precisely the kind of ability we value in a good collaborator: someone who knows an enormous number of patterns and is not afraid to leap from one domain to another.

If, for academics, this means creative introspection, for a CEO it means strategy: the same dynamic can be applied to dialogue with AI to explore organisational visions and market choices. As early as 1983, Donald Schön described the professional’s reflexivity as the capacity to converse with the situation.

Today we can say that a manager, if willing to push well beyond the first answers, can converse with their own strategy through an LLM and generate meaning that did not previously exist.

Beyond the tool: the extended mind

This changes AI’s role from a tool, however sophisticated, to a cognitive space. As Merleau-Ponty made clear, dialogue is not simply the transmission of information: it is a surplus of meaning – not pre-existing – that arises only within the relationship and could not otherwise have emerged. The key question is whether we can generate the same surplus by conversing with an interlocutor without a body.

With a language model, embodied reciprocity is absent: there is no other subject bringing their experience to the exchange. Yet it is precisely this neutrality that returns our words with statistical variations which, without intention, open up unexpected spaces of freedom. It is like speaking in front of a mirror that does not merely reflect, but slightly distorts the image, revealing details we had not noticed. It is in this discrepancy that value is produced: not a surplus of meaning in absolute terms – the prerogative of human reciprocity, as Geertz and Luhmann remind us – but new meaning for the leader engaged in dialogue.

Clark and Chalmers’s theory of the extended mind (1998) offers us a key to understanding this: tools and artefacts, if durably integrated into our cognitive processes, become part of our minds. Many people have experienced how a notepad or a search engine is already an extension of the thinking mind; in the same way, a chatbot can function as a cognitive extension, helping to reorganise thoughts, simulate scenarios and explore possibilities. It is not another mind sharing our world, but our own mind reworking itself. Consequently, the surplus is not automatic. It arises only through human intention: it is the leader who decides whether AI should remain a mere tool or become a true cognitive partner.

Suspending judgement and forming a habit

Because judgement and self-censorship are absent, AI can become an amplifier of the inner voice: in a board meeting, every idea is filtered through power, career considerations and status dynamics. With AI there is no embarrassment or reputational risk: we can experiment, take risks and even make mistakes with equanimity. It is like having a soundproof room in which to explore unspoken thoughts and hypotheses that are difficult to formulate.

Trust grows with habit. If a chatbot responds consistently and helpfully to the first question, then to the second and to many more increasingly uninhibited ones, the leader can begin to explore ideas with AI that they would not bring into a traditional setting. At first, they know they are receiving statistical responses; then, gradually, they forget that they are interacting with a system for creating strings of text and begin to converse with an interlocutor that functions as an extension of their own thinking. In this suspended space, the manager can articulate still-unformed intuitions, turn them into scenarios and bring forth strategic insights that would not come to light in an ordinary setting.

The surplus born of statistical lived experience

A structural limitation remains: the creation of genuinely new meaning, the surplus, always seems to require embodied experience. Yet AI brings with it a statistical lived experience: the fertile average of millions of texts that condense experiences and perspectives which were originally embodied. If this mass of knowledge is activated by the human interlocutor’s intention, it can produce not only well-formulated texts but also new situated meaning, insights that emerge within the particular conversation.

Perhaps this is where generative artificial intelligence truly comes into its own: not the generation of documents, but the generation of intelligence. This is what the most advanced literature on co-creativity (Eapen et al., 2023; Haase & Pokutta, 2024) is beginning to show: an ‘AI partner’ can expand human creativity and even generate autonomous creativity, producing results that are new and valuable. For a leader, this means having an interlocutor who does not replace thought but compels it to surpass itself. AI’s highest value lies not in what it knows, but in the way it obliges us to rework our own knowledge.

Since the use of AI as a strategic interlocutor is still in a pioneering phase, the conversations managers hold with it remain private. But experimental evidence also provides indirect confirmation of its potential. A study conducted by Boston Consulting Group (Dell’Acqua et al., 2023) involving 750 knowledge workers showed that teams supported by generative AI produced significantly more creative and higher-quality output in innovation activities: from generating new product ideas to developing business cases and defining launch plans. In other words, AI enhanced thought. It is a concrete example of how deep interaction with language models can create a cognitive surplus.

To be searching is already to be achieving

AI can become a tool for cognitive leadership only if the interaction is guided by a strong intention: to go beyond the first answers with the aim of creating something new, useful and marvellous. If you settle, you get sentences; if you aim higher, you open up a transformative semantic field.

The point is not to use AI, but to use it to build a strategy with substance. This requires a clear intention to achieve an outcome, and not stopping until a satisfying vision emerges. Failure can arise only from not having intended to achieve an extraordinary result and having broken off the dialogue too soon, not from a lack of answers that AI is capable of providing.

Milton Glaser (2012), perhaps the most influential designer of the twentieth century, the creator of the I ♥ NY, logo, shows that the creative miracle arises from the ability to follow the flow until we reach our destination. In the same way, those who use AI must not settle for answers; they must press on until they see authentic meaning emerge.

This is what the philosophy of language calls performative intention: the very act of declaring oneself in search creates the conditions for the existence of what one seeks. Intention is not neutral; it is a force that redirects perceptions, energies and decisions. The moment a leader declares to themselves what they are searching for – the creative miracle, for example, or an idea capable of opening up worlds – they change their own field of attention and creation: they perceive connections, opportunities and signals, and find ideas and words that would not otherwise have existed, bringing them into the dialogue with AI.

Meaning is not already present in the manager’s mind: it arises from acting in interaction with AI. This is the same principle Karl Weick describes through the concept of enactment: the organisation does not discover reality but constructs it by acting. Thus being in search is not a preliminary step, but an integral part of achievement.

From answers to vision

Imagine a manager who wants to launch a new line of organic products. They ask AI for some slogans that are not banal. They receive: The best technology is photosynthesis; Every ingredient has a family tree; Grown by time, not by chemistry. Original, but nothing extraordinary. On the other hand, this is the first response.

So they persist: ‘Give me slogans that no company in the sector would dare to use.’ AI responds: The caress that Nature returns with interest; When photosynthesis meets hygiene; The only luxury the Earth approves of. There is something there, but not enough if the intention is ambitious.

The manager pushes further: ‘How could I change the frame completely, to cast the concept in an entirely new light?’ AI draws on its statistical base and shifts the plane: Not cleaning but purification; Not a domestic gesture but a micro-act of ecosystem; Not a commodity but an act of resistance; Not a contemporary product but an archetype; A liturgical rite; An aesthetic symbol.

At this point the manager pushes again: ‘Rewrite them as religious texts and as titles of works of art.’ AI returns: Blessed are the small gestures, for great shall be their consequences; Green Renaissance. The first is discarded; the second becomes a revelation.

The manager no longer sees an advertising slogan, but a vision, a Green Renaissance for their organisation, a complete project in which the term ‘Renaissance’ is the fulcrum: it evokes the period when Italy knew how to fuse art and science, rediscover the ancient to project itself towards the new, and create beauty as a driver of progress. The ‘Green Renaissance’ then becomes a powerful metaphor: organic is no longer a product label, but a cultural framework capable of inspiring internal language, models of leadership and relations with the market.

The manager will return to creating slogans later. The miracle lies in having brought forth a new semantic field that can become a vision.

AI in slow motion

The manager who understands these transformative possibilities does not treat the machine as an assistant to whom tasks can be delegated, but as a true partner. The most impersonal technology humanity has built can become, for a leader, the most personal interlocutor with whom to realise their highest intentions and generate vision and strategy.

It is not a matter of speed or efficiency, nor even of persistence alone: it is the intention to achieve the miracle, to transform a brainstorming chat into a living strategy. It is this commitment that allows AI to cease being merely a tool and become a strategic-cognitive partner.

This article first appeared in Harvard Review Italia in December 2025: https://www.hbritalia.it/homepage/2025/12/17/news/utilizzare-lia-generativa-per-sviluppare-la-strategia-16393/ 

References

Clark, A., & Chalmers, D. (1998). The extended mind. Analysis, 58(1), 7–19. 

Dell’Acqua, F., McFowland III, E., Mollick, E., Lifshitz-Assaf, H., Kellogg, K., Rajendran, S., Krayer, L., & Candelon, F. (2023). Navigating the Jagged Technological Frontier: Field Experimental Evidence of the Effects of AI on Knowledge Worker Productivity and Quality. Harvard Business School Working Paper No. 24-013. 

Eapen, T. T., Faruquie, T., Mukherjee, A., Nori, H., Ribeiro, M. T., & Shokri, R. (2023). How Generative AI Can Augment Human Creativity. Harvard Business Review. 

Farri, E., & Rosani, G. (2025). How AI Can Help Managers Think Through Problems. Harvard Business Review.

Farri, E., & Rosani, G. (2025). HBR Guide to Generative AI for Managers. Harvard Business Review Press.

Geertz, C. (1973). The Interpretation of Cultures. New York: Basic Books. 

Glaser, M. (2012). In Search of the Miraculous: Or, One Thing Leads to Another. New York: Overlook Press / Abrams.

Haase, J., & Pokutta, S. (2024). Human–AI Co-Creativity: Synergies Between Human Ingenuity and AI Capabilities. arXiv preprint.

Luhmann, N. (1984). Soziale Systeme: Grundriß einer allgemeinen Theorie. Frankfurt am Main: Suhrkamp.

Merleau-Ponty, M. (1945). Phénoménologie de la perception. Paris: Gallimard.

Rockmore, D. (2025, August 9). What it’s like to brainstorm with a bot. The New Yorker. 

Schön, D. A. (1983). The Reflective Practitioner: How Professionals Think in Action. New York: Basic Books.

Turing, A. M. (1950). Computing machinery and intelligence. Mind, 59(236), 433–460. Oxford University Press.

Weick, K. E. (1979). The Social Psychology of Organizing (2nd ed.). Reading, MA: Addison-Wesley.