Why Communication and Critical Thinking Are the Most Underrated Capabilities in the Age of AI

AI prompting is both a communication and a critical-thinking skill.

And communication and critical thinking may be two of the most underestimated capabilities in organisations today.

The quality of what AI produces depends heavily on the quality of what a person can think through and explain. If an employee cannot define the problem, provide the right context, distinguish a useful answer from a plausible one or explain the outcome they are trying to achieve, AI cannot manufacture that clarity or critical judgement for them.

This is why organisations that treat AI capability as little more than technical training are missing something fundamental. Knowing where to click matters. Knowing what to ask, how to evaluate the response and how to communicate the resulting decision matters far more.

The organisations that gain the greatest value from AI will not simply have the most technically proficient workforce. They will have people who can think clearly, communicate precisely and apply judgement when the technology cannot.

Why We Still Call Essential Capabilities “Soft Skills”

Communication and critical thinking are often grouped under the heading of “soft skills,” as though they sit somewhere beside the real work rather than determining whether the real work gets done.

In practice, communication and critical thinking affect almost every organisational outcome. Strategy depends on leaders making sound choices and explaining their priorities. Execution depends on people understanding what is expected and questioning what does not make sense. Collaboration depends on teams exchanging and evaluating information without distortion. Customer experience depends on employees listening, interpreting and responding appropriately. Change depends on people making sense of what is happening and why.

AI adds another layer to this. Employees now have to communicate not only with other people, but with systems that respond literally to the context, constraints and instructions they are given.

The World Economic Forum’s Future of Jobs Report 2025 found that employers expect 39 percent of workers’ core skills to change by 2030. Technological literacy and AI-related skills are rising quickly, but so are distinctly human capabilities such as analytical thinking, resilience, leadership and social influence.

That combination matters. Technology changes what people can do. Communication and judgement determine whether they can use it responsibly and effectively.

A Good Prompt Begins Before Anyone Types a Word

Prompting is frequently presented as a formula: assign the AI a role, give it a task, add context and specify the format. These techniques are useful, but they are not the capability itself.

The capability begins before the prompt is written.

A person must first decide what problem they are actually trying to solve. They need to separate relevant information from noise, understand the audience, define what a good outcome looks like and recognise the constraints that must not be ignored. Only then can they communicate the request clearly enough for the AI to respond well.

This is no different from delegating work to another person. A manager who gives vague instructions, withholds essential context and changes the expected outcome halfway through cannot reasonably blame the employee for failing to read their mind. Yet many people interact with AI in exactly this way and conclude that the tool is unreliable when the response disappoints them.

Better prompting is not about finding magical words. It is about developing the discipline to think before asking.

Fluency Is Not the Same as Accuracy

AI can generate a polished response in seconds. That fluency creates a new organisational risk: language that sounds authoritative can be mistaken for thinking that is accurate. Communication shapes the request, but critical thinking determines whether the response deserves to be trusted.

Employees therefore need to do more than receive an answer. They must question it. What assumptions is it making? What information may be missing? Does the response reflect the organisation’s context, customers and regulatory environment? Is the source credible? Could bias or an error be hiding behind the confidence of the language?

Korn Ferry’s research on the skills an AI-ready workforce needs makes the same broader point: organisations require technical fluency, but they also need human skills that allow employees to work with AI thoughtfully. Critical thinking, adaptability, collaboration and communication are not secondary to AI adoption. They are what make responsible adoption possible.

Without those capabilities, AI may help an organisation produce more words, reports and presentations while quietly reducing the quality of the thinking behind them.

The Organisational Cost of Unchallenged Work

Poor communication and weak critical thinking rarely appear as a single line on a financial statement, which is one reason organisations underestimate them. Their cost is distributed across the business.

It appears in work that has to be redone because the brief was unclear. It appears in projects that slow down because decision rights were never explained. It appears when leaders believe a strategy has been understood because it was announced once. It appears in meetings where people leave with different interpretations of what was agreed. And it appears when an AI-generated recommendation is accepted without anyone questioning its logic or challenging its assumptions.

Korn Ferry’s current workforce-planning research reports that 43 percent of employees say their leaders are not aligned on how work should get done. That is not a minor engagement issue. Misalignment at leadership level travels through the organisation as contradictory priorities, duplicated effort and delayed decisions.

AI can accelerate this problem. When unclear instructions are fed into systems capable of producing work at scale, ambiguity is no longer confined to one conversation. It can be repeated across hundreds of outputs, workflows and customer interactions.

Speed does not correct confusion. It multiplies it.

What Working Effectively with AI Actually Requires

Organisations do not need every employee to become a prompt engineer. They need employees to communicate clearly, think critically and develop a more practical set of capabilities.

Problem definition. Before asking AI for an answer, employees need to frame the problem accurately. A badly framed problem can produce an impressive solution to the wrong question.

Contextual communication. AI needs relevant background, purpose, audience and constraints. Employees must know what context changes the quality of a decision and what information should never be entered into an external system.

Questioning and iteration. The first response is rarely the final one. Effective users probe assumptions, request alternatives, test logic and refine the conversation.

Evidence evaluation. Employees must distinguish between language that sounds credible and information that is supported by reliable evidence. This includes verifying facts and tracing important claims to their original sources.

Audience judgement. An AI-generated message may be technically correct and still be completely wrong for the person receiving it. Tone, timing, cultural context and organisational history remain human responsibilities.

Accountability. The person using the output remains responsible for the decision, recommendation or communication that follows. AI assistance does not transfer ownership.

These are communication capabilities, but they are also business capabilities. They influence quality, risk, trust and performance.

What I See Inside Organisations

Across leadership, culture and capability work, I often see organisations invest heavily in platforms while assuming employees will somehow develop the surrounding human skills by exposure.

A team is given access to an AI assistant and attends a short demonstration on its features. Within weeks, usage appears high. Employees are drafting emails, summarising documents and preparing presentations faster. On the surface, adoption looks successful.

Then the weaknesses begin to surface. Managers receive polished reports that do not answer the real question. Customer messages sound generic because nobody defined the audience properly. People accept inaccurate statements because checking the sources feels slower than generating the content. Junior employees become dependent on AI before they have developed enough subject knowledge to recognise when it is wrong.

The organisation has developed tool usage, but not capability.

That distinction is critical. Usage measures whether employees opened the system. Capability measures whether they can use it to improve the quality of work.

How Organisations Should Build These Capabilities

Traditional communication training often focuses on presentation skills, email etiquette or confidence when speaking. Critical-thinking development is frequently treated separately, if it is addressed at all. Those skills remain valuable, but AI requires organisations to connect them and widen the definition.

Teach people to frame problems. Give employees ambiguous, realistic business challenges and ask them to define the question before they use AI. The quality of the framing should be evaluated alongside the quality of the output.

Build verification into the workflow. Employees should know which claims require evidence, which decisions require human review and when subject-matter expertise must override a generated recommendation.

Use real work, not generic prompts. Capability develops when people practise with the decisions, documents, customer interactions and risks they encounter in their roles.

Develop managers as communication coaches. Managers should review how employees reached an answer, not simply whether the final document looks polished. Asking “What did you challenge?” may become as important as asking “What did you produce?”

Measure quality and judgement. Completion rates and platform logins tell leaders very little. Better measures include reduction in rework, accuracy, decision quality, customer outcomes and the employee’s ability to explain the reasoning behind the output.

The Capabilities AI Cannot Supply on Our Behalf

AI can improve a sentence, summarise a document and generate several possible answers. It cannot decide what matters most to an organisation without being given that context. It cannot take moral responsibility for a decision. It cannot understand the emotional weight of a message in the way the person receiving it will experience it. And it cannot create clarity that the user has not yet developed.

This is why the conversation about AI capability cannot be limited to technical proficiency, or assume that communication without critical thinking is enough.

The real advantage will come from combining technological fluency with the human capabilities organisations have always needed: communication, critical thinking, curiosity, adaptability and judgement.

AI may help people work faster. It may help them explore more alternatives and reduce time spent on routine work. But the quality of what it produces will still depend significantly on the quality of how people think, ask, evaluate and communicate.

Better communication leads to better prompts. Better prompts lead to better outputs. Better judgement turns those outputs into better organisational decisions.

In the age of AI, communication is not becoming less valuable because a machine can generate language. Critical thinking is not becoming less necessary because a machine can generate an answer.

Both are becoming more valuable because someone still has to determine whether that answer is sound and make the language mean something.