Communications professionals must push for a seat at the table when their organisations design AI systems, rather than being called in after implementation has created reputational damage. That was the central message from a panel at the PRovoke Media Asia-Pacific Summit in Singapore, featuring Amanda Koh, Edelman's head of client growth for Asia-Pacific, Jie Qi Lee, senior vice president at Edelman Intelligence, and Jess Tang, founder of mavic.ai.
Lee opened the session by asking attendees whether their organisations were investing more in AI this year than last year. Almost every hand went up. When she asked how many believed their organisations were genuinely ready, several hands came down.
Research cited by Lee shows that a significant percentage of companies plan to increase AI investment over the next three years, while more than half expect measurable returns within 12 months. Yet only a few have fully scaled AI across their organisations.
"The challenge is no longer accessibility," Lee said. "Almost everyone has access to AI. The challenge is that some companies are simply not ready yet."
From faster tasks to machine judgement
Much of the corporate conversation remains focused on selecting tools, training employees and completing existing tasks faster. Lee argued that organisations are moving towards a more consequential stage of adoption. The first phase involved using AI to accelerate individual tasks. The next will involve redesigning entire workflows and changing the division of labour between people and machines.
"Most of the execution is going to go to the machine, while humans move upstream into context-setting, approving, judging and deciding what we are going to let the machine determine," she said.
As AI evolves from a tool into an automator and eventually an agent, the central question shifts. "The question has changed from 'How can I make this faster?' to 'What are we allowing the machine to decide?'" Lee said.
That is where communications practitioners need to become directly involved. Tang said this distinction became clear as she moved from a communications background into building an AI platform alongside developers. "They see a text input and a text output with the correct word count, or a picture that has been created," she said. "I look at it and say, 'This is horrible. I cannot accept this.' Technically, it works, but it is not a piece of communication."
A system may complete its technical instructions while producing material that is culturally inappropriate, strategically weak or potentially damaging to the brand. "Only someone who has been in the industry and seen what happens on the ground understands what good looks like," Tang said. "Good is not simply whether something is nicely designed. It is whether it is culturally relevant and whether it puts the brand at risk."
Lee argued that AI transformation should not be left solely to technology teams. "Transforming AI practices within an organisation should not be the job of the communications team alone," she said. "But communications needs to come into the room and demand to be there."
"The AI teams are great at the technical work. They know what to do technically, but we are the ones who know what it takes to produce something good and what is important to retain in the process."
The APAC cultural gap
The issue is particularly acute in Asia-Pacific, where systems designed at a global headquarters may fail to reflect local languages, cultural expectations and political or social sensitivities. "Being in APAC, the cultural component is an extremely important part of the design process," Lee said. "Language and culture need to be taken into consideration when organisations review their AI processes. Too often that gets forgotten and we end up with a global AI protocol that does not apply locally."
Koh said the communications industry also needs to move beyond a narrow focus on productivity and content volume. "As communicators, we need to take a leap beyond productivity conversations and hundreds of content outputs," she said. "We need to think about how we add value to clients and stakeholders. If AI can produce the outputs and the content, where does communications stand and what value can we add?"
Part of that value lies in turning communications expertise into the instructions, standards and guardrails that shape how AI systems operate. "When practitioners are in the room, they can articulate all these invisible nuances," Tang said. "The opportunity with AI is that once you can articulate them, you can build them into the process because AI is a language model."
Otherwise, she warned, companies risk investing in platforms that generate material without producing meaningful communications outcomes. "You end up with a lot of software that gets things done without creating real communication outcomes," Tang said.
Measuring quantity, not quality
The panel also questioned whether companies have enough evidence to support claims that AI is already making their teams more productive. Many organisations measure adoption through usage, including how many employees have received training or how frequently approved platforms are accessed. Lee said those figures reveal little about the quality of the resulting work. "What we are measuring is quantity," she said. "The quality of the output has so far been missing."
Tang warned that an "invisible deskilling" is taking place as practitioners become accustomed to receiving immediate answers rather than learning how to develop and evaluate work themselves. "People are doing more, but the projects and business outcomes are not actually getting better," she said. "When you ask what they are looking for or what their criteria for good are, fewer people are able to articulate that."
This could prove particularly damaging for younger practitioners who begin using AI before they have learned how to research, structure an argument, write or recognise good work independently. "If we do not solve this now, it is going to move rapidly into the younger generation and we will end up with a talent shortage," Tang said.
Lee also highlighted the rise of "workslop", polished-looking AI material that lacks substance and forces the recipient to spend additional time deciphering or correcting it. "When someone produces paragraphs or slides using AI and gives them to you, they may look good on the surface," she said. "But when you read more deeply, the time you have to spend decoding what is being said is even greater."
She described the practice as disrespectful because it transfers responsibility from the person producing the work to the colleague, manager or client receiving it. "You are shifting the cognitive load of the creator onto the person who has to receive and review it," Lee said. "At that point, it is not helping productivity."
Communicators as human data scientists
For Tang, communications professionals could instead become "human data scientists", curating the experiences, emotions and contextual information that AI cannot independently acquire from online datasets. "A piece of documentation is not communication because communication has to communicate something," she said. "It needs to respect the person receiving it and it needs to be engaging for them."
AI can generate an output, but it still requires guidance about the brand, its stakeholders and the circumstances in which a particular tone or emotion is appropriate. "As human beings, we have access to data far beyond what exists on the internet," Tang said. "We have access to the way we grew up, our first heartbreak or the joy of achieving something. AI can mimic those emotions, but it probably cannot make a good decision about when to use them and when not to."
That ability to judge when and how to communicate may become more important as polished output becomes cheaper and easier to produce. Lee cited South Korean novelist Kim Ae-ran, who was asked what humans possess that AI does not. Her answer was "hesitation".
"Hesitation is the moment when you weigh whether something is the right thing to say to the right person at the right time," Lee said. "In our world, that hesitation has a name. We call it judgement."
AI platforms are designed to provide an answer, often confidently and immediately. Communicators, by contrast, are expected to recognise when an answer is inappropriate, when more context is required or when saying nothing may be the better decision. "AI compresses the cost of output and raises the value of judgement," Lee said. "We need to decide what deserves attention, where we should retain authority and what we are prepared to stand behind."
Why this matters for PR and communications professionals
The central question for organisations is not whether an AI system works as intended. It is whether the people designing it understand what good communication actually requires. As AI becomes embedded across corporate workflows, communicators cannot remain its final reviewers. They need to become some of its architects.
For practitioners, the practical implication is clear: develop the skills to articulate what good communication looks like in operational terms - cultural relevance, brand risk, audience context - and bring those standards into AI design conversations early. AI Learning Path for Public Relations Specialists offers a structured way to build that capability. For teams looking to embed these principles across their work, AI for PR & Communications resources provide a starting point for shifting from passive AI use to active governance.
The opportunity is not in producing more content faster. It is in defining the judgement, hesitation and cultural awareness that machines cannot supply on their own.
Your membership also unlocks:
