Three Lessons From Chatting About Strategy With ChatGPT
Three Lessons From Chatting About Strategy With ChatGPT
In MIT Sloan Management Review’s article “Three Lessons From Chatting About Strategy With ChatGPT,” authors Christian Stadler and Martin Reeves test whether generative AI can support the strategy-creation process. Published May 30, 2023, the article explores how ChatGPT performs when asked to help with strategic ideation, experimentation, evaluation, and storytelling.
For business leaders, the article offers a balanced message: generative AI can be a useful strategy assistant, but it is not a substitute for experienced strategic judgment. ChatGPT can generate ideas quickly, broaden the range of options, and help shape narratives, but it can also produce generic recommendations, miss proprietary context, and sound confident when it is wrong.
Executive summary for business leaders
Overarching theme: Generative AI can strengthen parts of the strategy process, but only when leaders use it as a thinking partner rather than an answer machine. Stadler and Reeves argue that ChatGPT’s greatest value lies in helping strategists expand possibilities, challenge conventional thinking, and draft strategic stories. Its limitations appear when strategy depends on company-specific knowledge, judgment under uncertainty, anomalies, analogies, and the ability to distinguish useful ideas from misleading ones.
The article identifies three core lessons:
Expect interesting input, not infallible recommendations. ChatGPT can generate useful ideas, but leaders should not treat its outputs as strategy.
Experienced strategists benefit most. The tool is most valuable in the hands of people who know how to ask good questions, challenge assumptions, and interpret the answers.
Strategists use different data differently. Human strategists rely heavily on proprietary knowledge, weak signals, analogies, anomalies, and context that AI may not have.
The practical lesson is clear: AI can accelerate strategy work, but it cannot replace strategic thinking.
Major takeaways
1. ChatGPT can support strategy creation, but not own it
Stadler and Reeves tested ChatGPT by giving it realistic strategic questions and using follow-up prompts to refine the responses. Their conclusion is not that AI can “do strategy” on its own, but that it can assist with several parts of the process.
Business implication: Leaders should view AI as a strategy co-pilot, not a strategy leader. Human judgment remains responsible for framing the problem, assessing options, and making decisions.
2. AI is useful for generating ideas quickly
One of ChatGPT’s strengths is speed. It can produce a large number of ideas, alternative business models, strategic options, and possible moves much faster than a traditional brainstorming process.
Business implication: Strategy teams can use AI to widen the top of the funnel. The value is not that every idea will be good, but that AI can help teams move beyond the first obvious answers.
3. Interesting input is not the same as reliable advice
The authors’ first lesson is to expect interesting input, not infallible recommendations. ChatGPT can sound persuasive, but its recommendations may be generic, incomplete, inaccurate, or poorly adapted to a company’s actual situation.
Business implication: Executives should resist the temptation to treat polished AI output as validated strategy. Fluency is not the same as insight.
4. AI can challenge conventional thinking — if prompted well
ChatGPT may initially produce conventional answers. However, with better follow-up questions, it can suggest alternative scenarios, disruptive moves, new business models, or different ways to frame the problem.
Business implication: The quality of AI-assisted strategy depends heavily on the quality of the prompts. Leaders should train teams to ask sharper, more specific, more challenging questions.
5. AI can help with strategic storytelling
The article notes that generative AI can support the storytelling side of strategy. It can help convert ideas into narratives, structure arguments, explain strategic logic, and draft communication for different audiences.
Business implication: AI can help leaders communicate strategy more clearly and quickly, especially when translating complex strategic choices into language employees, investors, or customers can understand.
6. AI struggles with company-specific context
Strategy is rarely based only on public information. It depends on internal capabilities, customer relationships, operating constraints, culture, talent, assets, economics, leadership appetite, and timing. AI may not know these details unless humans provide them.
Business implication: Strategy teams should not expect AI to understand the organization’s reality by default. Proprietary context must be supplied, protected, and interpreted carefully.
7. Experienced strategists benefit most from AI
The authors argue that experienced strategists are likely to benefit more than novices. Experienced leaders can recognize weak ideas, refine prompts, detect missing assumptions, and separate useful suggestions from misleading ones.
Business implication: AI does not eliminate the need for strategic capability. In many cases, it increases the value of human judgment because someone must evaluate the output.
8. AI can mislead inexperienced users
Because ChatGPT can produce confident and well-structured answers, less experienced users may overtrust it. They may not notice when the answer is shallow, generic, unsupported, or strategically flawed.
Business implication: Organizations should pair AI access with training in critical thinking, strategy fundamentals, evidence evaluation, and responsible AI use.
9. Strategists use data differently than AI does
Human strategists do not simply process large volumes of data. They often look for anomalies, weak signals, analogies, edge cases, contradictions, and patterns that do not fit the obvious answer. These are often the starting points for distinctive strategy.
Business implication: Competitive advantage often comes from seeing what others miss, not from summarizing what everyone already knows.
10. Proprietary insight remains a human advantage
AI tools trained on broad public knowledge tend to produce answers based on what is common, documented, or statistically likely. But distinctive strategy often depends on proprietary insight: customer data, internal know-how, operating constraints, leadership intent, market relationships, or emerging signals.
Business implication: Companies should combine AI-generated thinking with unique internal knowledge. The best strategy will come from blending broad AI-assisted exploration with proprietary human insight.
11. AI is better at expanding options than choosing among them
ChatGPT can help leaders develop more possibilities, but it is less reliable at deciding which option is right for a specific company in a specific context. The hard part of strategy is often not naming options; it is making trade-offs.
Business implication: Leaders should use AI to broaden strategic imagination, then use human judgment, data, debate, and governance to make choices.
12. AI can accelerate experimentation
The article tests whether ChatGPT can support experimentation and evaluation. AI can help generate hypotheses, identify potential tests, outline pilot structures, and suggest metrics.
Business implication: Strategy teams can use AI to move from abstract debate to testable hypotheses more quickly.
13. AI should be used iteratively
The authors did not simply ask one question and accept the first answer. They used follow-up prompts to challenge, refine, and improve the output. This back-and-forth process is where much of the value emerges.
Business implication: Leaders should teach employees to work with AI conversationally: ask, refine, challenge, redirect, specify, and test.
14. AI can help overcome blank-page strategy work
Strategy teams often struggle to begin. AI can provide a starting structure, list of options, scenario outline, or first draft that humans can then improve.
Business implication: AI may reduce the friction of starting strategic work, but it should not define the final answer.
15. Human judgment is still central to strategy
The article’s enduring message is that strategy is not just analysis. It requires judgment, imagination, context, experience, courage, and the ability to make choices under uncertainty. AI can support these activities, but it does not replace them.
Business implication: Companies should invest in both AI tools and strategic thinkers. One without the other will underperform.
Leadership talking points
Generative AI can help strategy teams generate ideas, structure options, and tell strategic stories.
AI output should be treated as input, not a recommendation.
The best results come from experienced strategists who know how to ask better questions and challenge the answers.
AI is strongest when expanding possibilities and weakest when proprietary context, judgment, and trade-offs matter most.
Strategy depends on company-specific insight, not only general knowledge.
Leaders should use AI to accelerate thinking, not outsource thinking.
The risk is not only that AI gives a bad answer; it is that a confident answer prevents deeper strategic work.
Reflection questions
Where in our strategy process could AI help us generate more options or test more assumptions?
Are we using AI as a thought partner or as a shortcut to answers?
Do our teams know how to challenge AI-generated recommendations?
What proprietary information would AI need to understand our real strategic context?
Are we training people to use AI critically, or simply giving them access to tools?
Where could AI help us create better strategic narratives for employees, customers, or investors?
Which strategic decisions require human judgment, experience, and debate no matter how good the AI output appears?
Are we preserving the development of strategic thinking while adopting AI tools?
Potential action items
Pilot AI-assisted strategy sessions for ideation, scenario planning, competitive analysis, and strategic storytelling.
Create prompt libraries for strategy teams, including prompts for alternative business models, market disruption, customer behavior, risk scenarios, and strategic trade-offs.
Require AI-generated strategy outputs to be reviewed against proprietary data, internal capabilities, customer insight, and financial realities.
Train leaders and strategy teams to evaluate AI output for generic thinking, unsupported assumptions, missing context, and false confidence.
Use AI to generate hypotheses and then test them through customer research, experiments, market data, and internal analysis.
Build a “human judgment checkpoint” into any AI-assisted strategy process.
Encourage teams to ask AI for contrarian views, alternative scenarios, second-order effects, and reasons a proposed strategy might fail.
Use AI to help draft strategy narratives, but have leaders refine the story for clarity, authenticity, and strategic conviction.
Protect sensitive proprietary information when using external AI tools.
Develop strategic-thinking capability alongside AI literacy so employees do not become overdependent on automated analysis.
Recommended similar articles
How AI Is Transforming Strategy Development — A strong McKinsey companion on how AI can assist with research, simulation, synthesis, and strategy communication.
Why Agentic AI Projects Fail — and How to Set Yours Up for Success — A useful HBR article on choosing the right AI use cases and avoiding poorly governed experimentation.
6 Ways AI Changed Business in 2024, According to Executives — A relevant HBR article on how AI is shifting executive priorities around data quality, governance, leadership, and enterprise value.
What I Learned from Daniel Kahneman — A McKinsey article on decision bias and why better decision processes matter in corporate strategy.
Mastering the Building Blocks of Strategy — A McKinsey article that complements this piece by grounding AI-assisted strategy in the fundamentals of strategic problem solving.
The Strategic Power of Hope — A helpful HBR companion for leaders who want to balance realism, possibility, and agency during AI-driven change.