The case, explained simply

BCG / Harvard · where AI helps

Match the tool to the task, rather than assuming every office task improves.

BCG / Harvard · where AI helps
Contextual photograph · Unsplash

Source period: September 2023 study · Source reviewed: 2026-11-10

THE WORK, MADE VISIBLEFrom information to a useful result
  1. Different work tasks
  2. With / without AI
  3. Blind quality review
  4. Choose where it helps

Explanatory workflow. Read the linked case or sector analysis for the evidence, conditions and limits behind each step.

What actually happened

The study involved 758 BCG consultants. Harvard’s summary reports faster work, higher rated performance and more completed tasks for tasks within the tested model’s capabilities. It describes a jagged boundary: tasks that look similar can respond differently to AI assistance.

Why this matters for adoption

A company should split a job into tasks before deciding where assistance belongs. Generating options, summarising supplied material and checking a complex argument are different activities. A tool can help with one and weaken another. This is why a single company-wide question—does AI work?—is less useful than asking which task improves, for whom, and with what checking.

THE CASE AT A GLANCE

Read a case like a decision-maker.

Do not stop at the headline. Follow the four things that make a case useful in a different organisation.

01The record

Named source: Harvard Business School · research summary

02The work

Which task actually changed?

03The proof

What was measured, by whom and in what setting?

04The boundary

What must your team test before it relies on this?

THE LESSON BEHIND THE CASE

What has to be true for this to help?

Performance can vary sharply between tasks that look similar to a non-specialist. An assistant may produce a persuasive answer outside its reliable range, making mistakes harder to notice. A company should therefore approve uses task by task rather than declaring one tool suitable for all knowledge work. Review the substance of a recommendation independently. Faster writing and a polished presentation are not substitutes for a sound decision.

A worked example — illustrative, not a reported result

Compare a straightforward summary with a recommendation that requires joining conflicting evidence. Score factual accuracy and reasoning separately from writing quality.

A PRACTICAL DECISION MAP

What your team needs to make this work

01 / PREPARE

Have this ready

Different task types with independently checked answers

02 / COMPARE

Look for this result

Reasoning and accuracy separately from presentation

03 / RECONSIDER

Pause if you see this

Polished output hides unreliable reasoning

An editorial guide for planning your own trial. Agree the required result and acceptable effort with the people who do the work.

What the evidence does not tell us

The study tested particular tasks and a 2023 model. It does not establish a universal productivity gain for all professional work or today’s systems.

What a Greek organisation can learn

  1. Choose one clearly defined task and the person responsible.
  2. Compare the complete work with the current process.
  3. Decide whether to extend, change or stop based on observed results.

How to test the idea in your organisation

Compare finished outputs without telling the reviewer which used AI. Include difficult tasks and record corrections. Do not evaluate only polished writing; check whether the conclusion is supported.

Before spending money

Ask for setup, integration, training, review, recurring charges and exit costs separately. Public case evidence is not a price quotation for your organisation. Agree a limited trial before committing to a wider rollout.

Read the original evidence

Harvard Business School · research summary ↗

The practical steps above are AIStrategy’s editorial suggestions. They are not reported actions or promised results from the organisation featured. Photography illustrates the subject unless its caption identifies the actual facility.