A finance colleague showed me a report recently featuring 20 AI tools deployed across the group, several thousand employees trained, and more than 40,000 hours saved last quarter. She looked pleased. Then she asked the harder question: What have we actually done differently as a business because of this?

We sat with that question for most of the afternoon. Neither of us had a clean answer.

AI has quickly moved from curiosity to expectation. A few years ago, the debate was whether to experiment with it. Today the conversation is about scale: tools deployed, employees trained, hours saved, processes automated.

These are useful measures, but they are not the same as business value.

A team can save three hours on a task and produce no better outcome. A company can introduce several AI tools without changing how work is organised. Employees can complete training programmes and continue using the technology only for low-value tasks.

The more useful question is not, “Are we using AI?” It is, “What is better because we are using it?” Faster decisions. Stronger reader experience. Better quality journalism. Lower costs. New revenue. More time on problems that require human judgment.

Therefore, every AI business case needs a talent case. It should explain not only what the technology will do, but how roles, skills, accountability, and ways of working will change around it.

I found it useful to hold this conversation in one place, through four connected questions related to capability, accountability, redesign and reinvestment, and enterprise value.

I have come to think of them as the CARE framework, because AI value depends less on how boldly an organisation deploys the technology and more on how carefully it thinks about the work, the people, and the outcome.

C: Capability

AI changes the value of distinct kinds of work. Routine processing and information synthesis become easier. Judgment, context, creativity, relationship building, problem solving, and the ability to question an AI-generated answer become more important, not less. Employees need to know when to rely on AI, when to challenge it, and when not to use it at all.

Not everyone moves up the same curve at the same speed. Some find the change energising and grow into the redesigned work quickly. Others find their current contribution shrinking, and the honest conversation about where they go next must happen earlier than most organisations schedule it. Pretending the transition is uniformly upward flattens what is a more difficult set of individual journeys.

Before approving an AI investment, organisations should be honest about the capabilities required to use it well. Can existing employees be developed? Do roles need to be redesigned? Are new specialists required? Are managers ready to lead teams where work is shared between people and technology?

Access to AI tools will soon be universal. Competitive advantage will come from how intelligently an organisation combines those tools with its distinctive human capabilities. Training people on a tool is only part of the job. The harder work is helping them apply it to real decisions, real audiences, and real consequences.

A: Accountability

When a person produces an outcome with the help of AI, who owns the result? If the recommendation is wrong or the analysis biased, where does responsibility sit? The answer cannot be with the system.

There needs to be clarity on who reviews the work, who makes the final call, and where human approval is essential. Oversight should scale with risk. An AI tool suggesting a headline sits in a vastly different risk category from an AI system influencing story selection, source attribution, or hiring decisions. Both need governance but not the same governance.

Performance measures have to evolve, too. When technology can generate output at scale, quantity becomes a weak signal of contribution. Quality of judgment, originality, responsible use of AI, and the ability to catch what the system missed become the better ones.

Organisations should also recognise the employees who improve AI-enabled workflows, spot better use cases, or prevent errors before they become expensive. These contributions sit outside traditional job descriptions but can create significant value.

R: Redesign and reinvestment

The temptation is to drop AI into an existing process and measure how much faster it runs. Sometimes that is enough. Often it is not.

Before automating a task, it is worth asking whether the task should exist in its current form at all. Making a bad process faster does not make it a good one. Work may need to be simplified, consolidated, reassigned, or eliminated altogether.

This is where managers make the difference. They decide which processes to challenge and which to leave alone. They decide whether saved capacity flows into deeper work or into another layer of reporting. They deal with the practical concerns their teams do not raise in formal meetings: worry about becoming less relevant, uncertainty about new expectations, quiet frustration with tools that do not work as promised. None of this is solved by a training module.

Every AI initiative should include a productivity dividend plan answering three questions:

1. What capacity will be released in time, cost, effort, or decision bandwidth?
2. Where will it be reinvested — whether into growth, innovation, quality, customer experience, or employee development?
3. How will the resulting value be measured?

At Bennett Coleman & Co. Ltd. (The Times of India), faster analysis or content processes create real value only when the released capacity is redirected into stronger insight, originality, verification, audience relevance, and commercial innovation.

Without such intentionality, AI makes organisations busier without making them better.

E: Enterprise value

The final test of AI is not whether the technology worked. It is whether the enterprise became stronger because of it.

Enterprise value shows up in specific business measures: Revenue generated or protected. Costs sustainably reduced. Decisions improved or accelerated. Audience experience strengthened. Product and campaign launches expedited. Quality and accuracy raised. Rework reduced. Innovation capacity increased. Employee effectiveness improved.

Not every AI initiative should produce all these outcomes. However, every AI initiative should be clear about which of these outcomes it is trying to produce, and how the organisation will know when it has arrived.

This moves the conversation beyond narrow cost reduction. AI is not primarily a mechanism for doing the same work with fewer people. Its greater potential lies in helping organisations improve the quality of work, create new possibilities, and deploy human capability against more valuable problems.

The takeaway

CARE is a simple test to hold in mind before, during, and after any AI investment:

• What capability is needed?
• Who is accountable?
• What is being redesigned and where the released capacity goes?
• What enterprise value the technology is expected to create?

The organisations benefiting most from AI will not be those deploying the most tools. They will be those that were clearest, from the start, about what should change, what should remain human, and what better outcome the technology was expected to create.

Banner photo: Adobe stock image by Aksana Kavaleuskaya.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .