AI Is Changing Consulting, but Judgment Is Becoming More Valuable

Artificial intelligence is transforming management consulting by driving the marginal cost of data synthesis, competitive research, and initial draft generation toward zero. However, the commercial value of advisory work is not evaporating. Instead, it is concentrating at the boundaries of human discernment: framing ambiguous problems, pressure-testing underlying assumptions, navigating organizational politics, and taking personal accountability for recommendations. In the evolving landscape of AI management consulting, advisory firms and corporate strategy teams are redesigning their organizational structures to reflect this operational shift.

This structural realignment is visible across major advisory firms. McKinsey & Company is growing its client-facing consulting roles by about 25% while non-client-facing roles have shrunk by roughly 25%, an approach its global managing partner calls "25 squared". The strategic rationale is straightforward: as generative models and specialized software agents automate back-office research and presentation drafting, the competitive differentiator shifts from sheer analytical volume to the high-touch judgment exercised directly in the boardroom.

The core economic reality of modern advisory work can be stated directly:

• Analytical execution is commoditized: Large language models, code interpreters, and automated research workflows can synthesize hundreds of financial filings, clean unformatted datasets, and draft preliminary slide decks in seconds.
• Structured problem-solving is premium: Deciding which strategic question to ask, establishing the boundaries of analysis, and defining the governing hypotheses cannot be automated by pattern-matching algorithms.
• Contextual judgment commands the margin: The client does not pay for twenty pages of generalized market commentary; they pay for a defensible decision path tailored to their capital constraints, operational realities, and risk appetite.

When the cost of producing analytical output collapses, the primary organizational bottleneck shifts from generation to evaluation. In traditional strategy engagements, junior consultants spent the bulk of their billable hours gathering data, formatting spreadsheets, and assembling presentations. In an AI-augmented environment, that raw output is available instantly. As advisory leaders at firms like EY have observed, when the cost of building analysis drops precipitously, determining exactly what to build becomes the critical determinant of strategic success.

This dynamic is accelerated by the widespread adoption of unsanctioned AI tools across corporate enterprises and consulting teams. Industry surveys indicate that a substantial share of enterprise workers use unapproved generative AI tools for daily tasks, often prioritizing speed over security and governance controls. When consultants and internal strategy teams rely on uncurated prompts, the resulting strategies frequently suffer from hallucinations, superficial consensus, and generic recommendations that ignore proprietary constraints.

Unverified AI analysis creates a dangerous illusion of analytical rigor. Because large language models generate fluent, confident prose and structured tables on demand, teams risk mistaking grammatical coherence for strategic validity. Human judgment is required to verify the chain of custody behind every data point, challenge unstated premises, and ensure that recommendations reflect genuine market dynamics rather than statistical averages.

1. Volume inflation: AI tools flood decision-makers with plausible-looking reports, increasing cognitive load without increasing decision clarity.
2. Erosion of source verification: Unsanctioned AI tools frequently draw on outdated or unverified training data without transparent citations.
3. Convergence of ideas: Because generative models predict the most probable sequence of words, uncritical reliance on AI produces homogenized, middle-of-the-road strategic plans that fail to create distinct competitive advantage.

To navigate the shift from volume-driven deliverables to high-conviction decision-making, advisory professionals must operate within a structured evaluation hierarchy. The Consulting Judgment Premium framework defines the seven progressive layers where human expertise adds non-substitutable value over algorithmic computation.

1. 1. Problem Framing: Structuring ambiguous, messy business situations into crisp, testable problem statements using rigorous strategic problem framing and MECE problem solving architectures.
2. 2. Evidence Quality: Auditing raw data, identifying methodological bias, verifying source provenance, and separating genuine proprietary signals from public noise.
3. 3. Assumption Testing: Explicitly isolating unstated operational, economic, or regulatory assumptions that underpin a business model and subjecting them to rigorous sensitivity stress-testing.
4. 4. Trade-Off Judgment: Evaluating mutually exclusive strategic choices across capital allocation, risk tolerance, and opportunity cost where no mathematical optimum exists.
5. 5. Stakeholder Context: Reading organizational power dynamics, cultural resistance, leadership incentives, and change readiness to determine what is politically and operationally viable.
6. 6. Recommendation Synthesis: Converting multifaceted analytical findings into a clear, actionable, board-ready verdict that directly answers the governing question.
7. 7. Accountability and Ownership: Standing behind the recommendation, taking personal professional responsibility for implementation risks, and guiding executive leadership through the execution journey.

While advanced reasoning engines can propose theoretical alternatives and summarize trade-offs, they possess neither skin in the game nor contextual intuition. The final layers of the framework - trade-off resolution, stakeholder alignment, and executive accountability - remain inherently human responsibilities that boards cannot outsource to an algorithm.

Understanding where AI accelerates work and where human judgment is non-negotiable requires examining specific consulting workflows. A joint experiment by Boston Consulting Group and Harvard's Digital Data Design Institute (now the Harvard Business School AI Institute) evaluated the performance of 758 consultants, roughly 7% of BCG's individual-contributor workforce. It produced the term "jagged technological frontier": some tasks are easily done by AI, while it cannot do others that are similar in difficulty, which is why companies must learn when the tool helps and when it does not.

In a follow-up BCG study presented with the Harvard Business School AI Institute, consultants with no programming experience using ChatGPT achieved 17% of what a data scientist could accomplish within a 90-minute data cleaning task. While this demonstrates the power of AI as an analytical accelerator, it also highlights the substantial gap in deep contextual understanding, edge-case detection, and statistical validity that only trained human professionals provide.

As the comparison illustrates, delegating execution tasks to AI delivers massive velocity gains, but delegating judgment creates severe organizational risk. High-performing strategy teams treat AI as a junior research accelerator while maintaining strict human-in-the-loop checkpoints across the entire analytical pipeline.

As organizations adopt generative AI across daily workflows, they frequently encounter a subtle but systemic organizational pathology: distributed de-skilling. In a recent survey, half of senior executives said they are already seeing the collective judgment of their workforce erode, and 60% said this "de-skilling" poses a significant threat to their organizations over the next few years. A Boston Consulting Group study of 70 senior executives found that the thinking skills leaders prize most, including judgment, problem framing, and original analysis, are eroding fastest as AI saturates daily work.

Distributed de-skilling occurs when teams offload fundamental cognitive tasks to automated models, bypassing the iterative problem-solving repetitions that develop professional discernment. In management consulting, junior analysts historically built deep strategic intuition by manually reconciling messy balance sheets, conducting primary expert interviews, and structuring issue trees from scratch. When AI automates these entry-level tasks end-to-end, the traditional talent on-ramp breaks down.

• Uncritical acceptance of model outputs: as AI access spreads, the volume of mediocre, "good enough" work is climbing, and teams accept AI-generated analyses without interrogating the underlying logic.
• Dilution of personal ownership: When recommendations underperform, team members deflect accountability by pointing to algorithmically generated proposals.
• Cognitive atrophy in problem framing: Teams lose the ability to define root causes independently, relying on generic chatbot prompts that yield homogenized strategies.
• Erosion of apprenticeship models: Senior leaders spend less time coaching juniors through the mechanics of analytical reasoning, reducing the transfer of tacit industry knowledge.

To protect institutional capability, consulting leaders must practice radical transparency regarding technical realities and model limitations. Building resilient advisory capabilities requires intentional workflow governance that separates automated data gathering from core cognitive development.

Embedding the Consulting Judgment Premium into daily strategy engagements requires moving beyond ad-hoc prompting to disciplined, human-led workflow architecture. Strategy teams must establish clear protocol boundaries that dictate when AI is deployed for acceleration and when human deliberation is mandatory.

The Human-Led Workflow Protocol

High-performing strategy practitioners implement a sequential gating process that ensures human problem framing always precedes algorithmic synthesis:

1. Step 1: Human-Led Problem Scoping: Define the core strategic question, stakeholder constraints, and governing hypothesis tree before opening an AI interface.
2. Step 2: AI-Accelerated Evidence Gathering: Deploy approved enterprise models to ingest research papers, summarize public financial statements, and structure data tables.
3. Step 3: Verification and Provenance Audit: Trace every cited metric back to its primary source document, eliminating hallucinated figures and unverified assumptions.
4. Step 4: Human-in-the-Loop Stress Testing: Subject the synthesized findings to pre-mortem analysis, alternative scenario simulations, and competitive counter-moves.
5. Step 5: Stakeholder and Political Calibration: Adjust the final strategic options to match the client organization's real risk appetite, leadership incentives, and change readiness.

Strategy Workflow Checklist

• Are strict AI-off zones established for ethical evaluations, core executive trade-offs, and final recommendation synthesis?
• Is every quantitative data point in the final presentation backed by a verified, human-checked primary source?
• Did the team conduct an independent assumption-testing session to identify what must be true for the recommendation to succeed?
• Does the recommendation clearly articulate the trade-offs and opportunity costs of the chosen path versus mutually exclusive alternatives?
• Has a single project leader taken personal accountability for the strategic guidance, independent of the tools used to produce it?

Managing strategic decisions in an AI-accelerated business environment requires a structured bridge between analytical inputs and executive execution. Decisity serves as a dedicated strategy consulting and management decision-support knowledge hub, empowering leadership teams to transform complex, ambiguous challenges into defensible board-ready choices.

Rather than functioning as a black-box analytical tool, Decisity grounds the advisory process in a rigorous, transparent chain of reasoning: moving systematically from problem framing and evidence synthesis to alternative generation, trade-off evaluation, strategic decision-making, and strategy execution framework delivery. By enforcing disciplined problem decomposition and transparent assumption auditing, Decisity ensures that enterprise teams harness the speed of modern AI without compromising on human strategic judgment.

• Harvard Business School AI Institute: navigating the jagged technological frontier
• Harvard Digital Data Design Institute: a large-scale BCG skill-development experiment
• BCG: when everyone uses AI, companies risk losing critical skills
• McKinsey's global managing partner on AI reshaping the firm's workforce