AI Won’t Replace Marketers—But It Will Replace Marketing-as-Usual
by Gemma Calvert
Background
If you work in marketing, the ground under your feet is moving—fast. In a live panel at NTU, three industry leaders cut through the hype: Krishnan Menon (Managing Director, R/GA), James Breeze (CEO, Research Network), and Melvyn Tan (Entrepreneur & Co-founder, Befinity AI). Their verdict? Generative AI is rewriting the playbook—from research and production to capability building and legal risk. The marketers who thrive will be those who go upstream: inventing the brief, shaping the strategy, and designing the future—while machines accelerate the rest.
1) Use AI as a co-pilot, not a crutch
For James Breeze, AI has become an “always-on companion” that collapses the time from question to insight.
“I use it ten times an hour,” he said. “For ideas, quick research, and faster workflows. The change to my business management is insane.”
But he drew a critical line: if you’ve never done the thing in the real world—written a brief, built a website, run a user test—you can’t judge if the AI’s output is good or garbage.
“If people haven’t done the task before, they can’t tell if the recommendations are right or wrong,” he warned. “That’s why great marketers will keep winning: they know what ‘good’ looks like—and use AI to get there faster.”
Takeaway: AI multiplies judgment. If you lack foundations, you’ll multiply confusion. Make your base craft non-negotiable.
2) Prompt engineering is hot—and short-lived
Melvyn Tan calls prompt engineering “the most important communication skill to learn right now”—with a twist.
“We’re in the MS-DOS era of AI,” he said. “You need to know how to ask, frame and scaffold to avoid generic output. But this skill has the shortest lifespan ever.”
Why? Because custom assistants and agentic systems will soon absorb the hard parts of prompting.
“You’ll configure a Custom GPT or Intelligent Assistant with deep system prompts and chains,” Melvyn explained. “End users will simply say, ‘Give me a video script for this product,’ and get nuanced output. The complexity gets abstracted away.”
Takeaway: Learn prompting to think clearly and communicate constraints. But invest more in problem framing, domain expertise, and workflow design—they endure.
3) Personalisation at scale beats boxing people in
Both Melvyn and James challenged the old model of slotting humans into fixed “types.”
“Boxes helped when we couldn’t scale personalisation,” said Melvyn. “Now we can design highly personalised strategies—learning plans for kids, interventions for consumers—and let systems recalibrateas new data arrives.”
James has already re-architected research:
“We’re running qual at scale with AI interviewing—participants respond on video or voice, without filters. You train the model how to synthesise, and you get richer, faster insight than forcing people into survey boxes.”
Takeaway: AI lets us meet people as people. Move from segments to signals, from averages to adaptive experiences.
4) Legality and trust: the elephant in the (creative) room
Krishnan Menon drew a bright red line through the industry’s current behaviour.
“We keep saying ‘AI,’ but we usually mean generative AI,” he began. “And much of how it’s used in marketing today is legally unusable. If the output borrows from rights-controlled content or data you don’t own, brands can’t deploy it.”
He shared a hard truth from the front lines: agencies are building exciting pilots clients won’t ship—not because the work isn’t good, but because the rights and perception risks are too high.
“A global food brand refused to use stunning gen-AI imagery,” Krishnan said. “Food photography already faces claims of misleading. Add ‘AI-generated’ to that and you create a trust problem.”
Takeaway: Move AI upstream: design systems that start from owned data, owned assets, and clear consent. If you can’t ship it, it’s a demo—not a strategy.
5) Go upstream: strategy, invention, and design thinking
Krishnan’s career advice for students and practitioners was blunt:
“AI will replace the downstream. It won’t replace the upstream. Pursue thinking jobs—creating, designing, deciding. Humanity brings the vision of the future; generative AI brings signals from the past.”
That means shaping the problems worth solving, defining “better,” and architecting workflows where humans and machines compound one another.
Takeaway: If your work is easily specified after the brief is set, it’s vulnerable. If your value is setting the brief—and translating vision into repeatable systems—it’s about to scale.
6) Stay current—or get blindsided
James offered a war-story with a sting. He quoted ~AUD 50k for a UX research project to prove a clunky purchase flow was hurting conversion. Then he tested a new Gemini capability on a sandbox.
“In 20 seconds I had evidence backing the Australian team’s position,” he said. “The lesson: you have to be across what’s new—every single day.”
Takeaway: Build a discovery habit—scan weekly updates, maintain a peer network, and pressure-test old processes against new tools. The cost of ignorance is now immediate.
7) What to do on Monday
- Audit legality: inventory where your AI outputs come from. Shift to owned/cleared datasets and assets.
- Move upstream: carve out time for brief invention, not just content production.
- Design assistants, not prompts: capture your best prompts as reusable assistants tied to your workflow and guardrails.
- Upgrade judgment: pair juniors with veterans—let novices learn the craft so AI doesn’t lead them astray
- Measure speed-to-signal: track how fast teams move from question → insight → decision using AI. Reward cycle time, not slide counts.
- Institutionalise learning: set up a shared AI changelog and fortnightly show-and-tell. If it’s not in the workflow, it’s lost.
Final word
The marketers who win won’t be those who fight the machines. They’ll be the ones who design how work is done, where human judgment and machine acceleration meet. As Krishnan put it: the past is data; the future is ours to imagine.
Panel
- Krishnan Menon, Managing Director, R/GA
- James Breeze, CEO, Research Network
- Melvyn Tan, Co-founder, Befinity AI
- Moderator: Prof. Gemma Calvert, Nanyang Technological University (NTU)