Generative AI can help organisations communicate faster and at greater scale. But unless every message is grounded in evidence, operational reality and human accountability, polished content can weaken rather than strengthen customer trust.
The debate triggered by reports of AI-like phrasing in speeches at the 81st UN General Assembly is easy to reduce to a question of authorship: Did a machine write these words?
That is not the most important question.
The more consequential question is whether audiences can still recognise genuine conviction, institutional responsibility and lived understanding in the words they hear from leaders and brands.
For customer-experience leaders, this is not a distant debate about politics or public speaking. It is an immediate business issue. Organizations already use generative AI to draft marketing copy, personalize offers, summarize service interactions, create knowledge articles, translate customer communications and power conversational experiences. Used with judgment, it can make brands faster, more relevant and more useful. Used without it, AI can make a brand sound polished while revealing that no one has thought deeply about the customer’s situation.
The real risk is not that customers will detect AI. The real risk is that they will detect a lack of thought.
The UN example: polished language, weakened signal
According to GPTZero, its analysis of 768 speeches delivered across 12 UN General Assembly meetings classified 107 speeches as AI-generated. The company reported that these speeches contained 57,729 of the 348,853 words analysed, or 16.5 percent of the total. It also identified similarities in language used by leaders from different countries, particularly in passages about trust, declarations, commitments and representation. These findings indicate patterns in the text; they do not, by themselves, establish how any speech was produced.
None of this proves that a leader who used a familiar phrase relied on AI. Diplomatic language has always been repetitive. Leaders responding to the same theme will naturally share vocabulary. But the episode matters because it captures a growing audience reaction: when language becomes too familiar, too frictionless and too interchangeable, people begin to question whether there is a real person—and a real commitment—behind it.
At the UN, the cost is diplomatic credibility. In business, the cost is customer trust.
A customer does not need a detector to make that judgment. They experience it directly.
They experience it when a bank sends an elegant fraud-warning email that does not tell them what action to take. They experience it when an insurer expresses sympathy after an accident but makes the claims process harder to navigate. They experience it when a retailer calls an offer “personalized” despite ignoring what they bought yesterday. And they experience it when a chatbot apologizes fluently but cannot resolve a simple, urgent issue.
In these moments, the problem is not artificial intelligence. The problem is artificial care.
Content is part of the experience
Many organizations still treat content as a marketing asset: a campaign, email, social post, landing page or article. In reality, content is part of the customer experience architecture.
It sets expectations before a purchase. It explains a decision during onboarding. It reduces effort when something goes wrong. It shapes whether customers believe a promise has been kept.
That means AI-assisted content cannot be evaluated only on speed, volume or grammatical quality. It must be evaluated on whether it helps customers understand, decide and act.
Consider a customer whose card is declined while travelling.
An AI-enabled service layer can be genuinely valuable. It can recognise the likely reason for the decline, verify the customer securely, provide a clear next step, surface the right self-service option and give a credible timeframe for resolution. It may prevent a call, reduce anxiety and restore confidence within minutes.
But now consider the alternative: a beautifully written message that says, “We understand how frustrating this must be,” before sending the customer through three channels, repeating authentication details and offering no resolution path.
That is not empathy.
It is automated theatre.
The distinction matters because customers judge brands by the usefulness of the experience, not by the polish of the language around it.
The authenticity gap
Generative AI gives organizations the ability to create more content at lower cost than ever before. That is valuable—but it also creates an authenticity gap.
The faster content is created, the easier it becomes to skip the parts that make it credible: customer insight, subject-matter expertise, evidence, local nuance, operational ownership and a clear point of view.
What AI can improve | What goes wrong without human judgment |
Campaign turnaround time | Messages are published before claims, implications and relevance are properly tested |
Personalisation at scale | Customers receive offers or advice based on incomplete, inaccurate or insensitive context |
Brand consistency | Every message adopts the same sterile, overly polished voice |
Customer-service productivity | Automation acknowledges problems without solving them |
Localisation | Translation loses cultural meaning, context or regulatory nuance |
Content volume | Brands create more noise while producing less distinction |
Executive communication | Leadership language becomes generic and disconnected from real action |
This is why the best test of AI-assisted content is not “Does it sound human?”
A more useful question is: Could any competitor have said this?
If the answer is yes, the content is likely generic. If the message works after replacing your organization’s name with another brand’s, it is not yet carrying a distinctive point of view.
Authenticity comes from proprietary truth: what an organization knows because of its customers, products, employees, performance, local context and choices. AI can help express that truth. It cannot manufacture it reliably.
From content generation to accountable communication
The opportunity is not to eliminate AI from the content process. It is to use it with clearer boundaries.
The way forward is that organisations should move from AI-generated content to AI-enabled human communication.
The difference is simple.
AI-generated content begins with an output requirement: create 20 social posts, write a CEO note, produce a product launch campaign, draft an apology email.
AI-enabled human communication begins with a customer and a responsibility: What does this person need to know? What is the organization actually prepared to promise? What evidence supports the claim? What will happen if the customer follows the advice? Who owns the outcome if the answer is wrong?
That shift changes the role of both technology and people.
1. Use AI to accelerate the work, not replace the judgment
AI is excellent at producing drafts, adapting approved messages across channels, suggesting alternatives, simplifying complex language, identifying gaps and helping teams work faster.
But it should not be left to invent product claims, interpret regulation, fabricate customer sentiment, make unverified market assertions or simulate executive conviction.
For high-stakes communication—financial, health, legal, crisis, complaint-resolution, public-policy or executive messaging—a named human owner must remain accountable for accuracy, tone, impact and follow-through.
AI can propose the words. A person must own the promise.
2. Ground every message in evidence customers can feel
The strongest content does not sound impressive because it uses elevated language. It feels credible because it reflects something real.
That might be:
A recurring customer problem identified in call transcripts or journey research.
A clear product or service capability that can be demonstrated.
A process improvement that genuinely reduces customer effort.
A local insight about language, culture, accessibility or regulation.
An employee’s frontline expertise.
A customer story used with appropriate permission.
A candid acknowledgement of a limitation, along with what the organization is doing about it.
For example, “We are committed to helping customers during difficult moments” is a familiar promise. A more meaningful statement would explain what changes in the experience: faster claims status updates, a dedicated escalation route, proactive assistance after a disruption or clearer documentation requirements.
Specificity is not merely a writing choice. It is evidence of operational seriousness.
3. Design the experience before drafting the message
Too many organizations use AI to improve the wording of a broken journey.
That is backwards.
If the underlying process requires customers to repeat information, switch channels or wait without clarity, more empathetic copy will not solve the problem. It may even make the gap between promise and delivery more obvious.
The right sequence is:
Identify the customer task and emotional stakes.
Fix the decision, workflow, hand-off or resolution path where needed.
Define the information the customer genuinely needs.
Use AI to make that information clearer, more relevant and easier to access.
Measure whether customers completed the task with less effort and greater confidence.
This is where AI creates real customer-experience value: not as a content factory, but as an enabler of better decisions and more useful interactions.
4. Measure trust alongside productivity
Content teams will understandably measure speed, output and cost. Those measures should remain on the dashboard. But they are not enough.
An AI-enabled content program should also monitor:
Customer comprehension and task-completion rates.
Repeat contacts after an automated response.
Escalation from self-service or chatbot interactions to human support.
Complaint rates related to unclear, inaccurate or misleading communication.
Content accuracy, compliance and substantiation failures.
Sentiment in customer feedback and service conversations.
Brand-distinctiveness measures.
The gap between customer expectation created by a message and the experience actually delivered.
A reduced cost per asset is not a success if it increases confusion, repeat contact or distrust.
The most valuable measure is simple: did this communication make it easier for the customer to move forward?
5. Be transparent where it matters
Not every use of AI needs a disclosure label. A grammar check, headline variation or translation draft is not the same as an automated decision or an AI-led customer conversation.
But when AI materially shapes an interaction, customers should be treated with respect. They should know when they are engaging with an automated system, understand what it can and cannot do, have a practical way to reach a person when needed and know who is accountable when something goes wrong.
Transparency works best when it is built into the design of the journey—not buried in a policy page.
A new premium on human judgment
The future will not reward organizations simply for using AI. AI capability will become widespread. The differentiator will be whether a brand can combine AI’s speed with human context, empathy, expertise and accountability.
This is particularly important in India, where customer experiences often span multiple languages, varying levels of digital confidence, deeply personal financial and family decisions, and high expectations for responsive service. A generic response may be technically correct and still fail the customer because it does not account for context.
That is why human judgment becomes more—not less—valuable in an AI-rich environment.
The strongest organisations will use AI to remove unnecessary effort from content work and customer journeys. They will reserve human attention for the moments that require interpretation, care, judgment and responsibility. They will ensure that messages are rooted in evidence, not merely optimized for engagement. And they will recognize that authenticity is not achieved by making automated content appear human.
It is achieved by making sure there is a real organization behind the content—one that understands the customer’s situation, can explain its choices and is prepared to deliver on its words.
The question content leaders should ask
Before publishing an AI-assisted message, content and CX leaders should ask:
If a customer learned that AI helped create this communication, would that discovery reduce or reinforce their trust in us?
If the answer is “reduce,” the solution is not to hide the use of AI.
The solution is to improve the evidence, judgment, transparency and accountability behind the message.
Because in a world where any organisation can generate eloquent content in seconds, eloquence is no longer scarce.
Trust is.
