AI can now turn five days of workshop discussions into a presentation in about 30 minutes. But does producing work faster necessarily make someone a better performer? For Kameshwari Rao, Global Chief People Officer at Publicis Sapient, the distinction matters as organisations rethink how they measure performance, identify talent and make promotion decisions.
In an interaction with People Matters, Rao explains why AI is changing the expectations attached to jobs without making traditional measures of merit obsolete. Employees must still deliver results, she says, but the ability to question assumptions, choose the right solution and take responsibility for AI-assisted work is becoming increasingly important.
The challenge for employers is to recognise these capabilities without allowing output volumes, access to technology or subjective managerial assessments to determine who gets ahead.
When producing more is no longer the same as performing better
Rao's own experience illustrates how AI can change the nature of work, not simply accelerate it. Following a five-day workshop, her team used digitally captured conversations to create a five-slide presentation in approximately 30 minutes, a task that would previously have required several people to review the material and build the slides.
The time saved was significant, but the process still demanded human intervention. AI-generated content needed to be reviewed and corrected, and the team had to define the desired outcome before using the technology effectively.
“So AI didn't just make the work faster. It changed how we approached the work itself,” Rao says.
For organisations, this shift complicates a familiar question: how should performance be judged when employees can produce substantially more with AI?
Rao cautions against rewarding volume without examining value. Clients are already factoring AI-enabled efficiencies into expectations around delivery and cost, potentially making yesterday's strong performance today's baseline.
“The more meaningful distinction is the value of the work itself,” she says, adding that she would rather see an employee find a better solution to a client's problem than simply produce more deliverables.
She acknowledges that organisations have not yet worked out precisely how to assess these differences. Managers must look beyond productivity dashboards to understand the contribution behind the numbers.
The real test begins when employees question AI
If AI can generate multiple solutions, an employee's contribution cannot be judged solely by the material produced. Rao believes the decisions behind the output offer a more useful indication of capability.
She is not convinced organisations need to calculate how much of a deliverable came from a human and how much from AI. Instead, she wants employees to demonstrate the knowledge required to evaluate the results and take responsibility for what they use.
“And sometimes the most valuable contribution is deciding not to use what AI has given you.”
This places a premium on domain knowledge and critical judgement. Employees must conduct their research, recognise unreliable answers and understand why a particular approach makes sense for the business.
The distinction is particularly relevant as AI tools become more accessible across functions. Greater access can improve productivity, but it does not automatically establish that an employee has made the best decision or created the greatest value.
For Rao, the question is less about which tool someone used and more about whether the individual can explain the choices that shaped the outcome.
Why a polished deliverable may reveal very little about potential
Rao offers another example of why organisations need to look beyond finished work when assessing talent. During a meeting to review a piece of work, her team concluded that a presentation was not needed at all.
An employee could have spent hours creating an impressive deck, potentially with AI. Yet recognising what the audience actually needed was the more valuable decision.
“That's a different kind of capability. It requires perspective. It also requires the confidence to challenge what appears to be the obvious answer.”
For leaders identifying high-potential employees, Rao highlights two qualities: the ability to read a situation and make sound decisions, and learning agility, or the willingness to adapt as work changes.
She suggests managers ask employees why they chose one approach over another, rather than relying only on the final deliverable. The explanation can reveal how someone evaluates alternatives, understands the problem and thinks through consequences.
AI may also bring forward expectations traditionally associated with more experienced professionals. Junior employees can now develop several possible approaches quickly, leaving them to determine which option best addresses a client's challenge and make a convincing case for their choice.
“That's a level of business acumen we might previously have expected from someone much further along in their career,” Rao says.
Can promotions remain fair when access to AI differs?
The challenge extends beyond performance reviews to compensation, promotion and career progression. Employees use different tools, work in different functions and face different expectations, making a single productivity benchmark inadequate.
Rao argues that fairness does not require identical treatment. Organisations should provide access to foundational AI learning and tools, particularly for early-career employees, so people are not disadvantaged because they lack opportunities to experiment.
Advanced access can then reflect the requirements of a role. Product engineering, for instance, may demand different AI capabilities from operational work, with corresponding differences in career paths and promotion criteria.
“We cannot expect everybody to demonstrate the same kind of productivity,” she says.
There is also the question of subjectivity. Rao notes that performance evaluations have always involved managerial judgement, even when organisations translate assessments into numerical ratings. AI has not created this issue, but its growing role has made it more visible.
At Publicis Sapient, she points to a process in which employees can nominate themselves for promotion, with advocacy and sponsorship also possible beyond their immediate manager. Bringing more perspectives into the process, she says, can reduce dependence on one person's assessment, although it cannot eliminate subjectivity entirely.
The broader challenge is to ensure that performance systems recognise the demands of different roles while assessing employees on meaningful contributions rather than raw output alone.
What meritocracy could look like next
Rao does not believe organisations need to discard meritocracy because AI has become part of everyday work. The expectation to deliver outcomes remains, but the bar against which those outcomes are judged is changing.
Managers will need to understand not only what an employee has delivered, but also how they reached the result, why they selected a particular solution and whether it addressed the actual problem. The same evaluation approach will not necessarily work across every role.
Rao also acknowledges that organisations are still working out how best to assess performance and potential in an AI-enabled workplace. As the technology evolves, their approaches to measuring merit will need to evolve as well.
Her argument ultimately returns to a distinction that AI makes harder to ignore: generating options is not the same as knowing which option matters.
“We're moving towards a world where how someone thinks will matter just as much as what they deliver.”
