AI didn’t replace insight. It changed where it comes from

AI is everywhere in marketing right now. Faster analysis. More outputs. More versions. More predictions.

And yet, when I facilitated a recent CMA Insights Council discussion, what struck me wasn’t how much more data we now have. It was how much more judgement is required to use it well.

The real shift isn’t speed or automation. It’s where insight comes from, how decisions get made, and what responsibility marketers now carry when AI enters the process.

That conversation surfaced three tensions Canadian marketers are actively navigating as AI becomes embedded in insight, planning and execution. None of them have simple answers. All of them require intentional leadership.

From more data to better decisions

AI has dramatically expanded our ability to identify patterns, predict behaviour and surface insights at scale. Tasks that once took weeks now take minutes.

But speed alone doesn’t guarantee better decisions.

As one Council member put it during our discussion, “AI is great at patterns, but not meaning.” It doesn’t understand brand nuance. It doesn’t carry lived experience. And it doesn’t know when not to act on an insight.

That gap between output and interpretation is where marketers now add the most value.

Insight is no longer just about what the data says. It’s about deciding:

• What matters,
• What’s ethical,
• What’s useful, and
• What aligns with brand values and customer trust.

Tension one: personalization vs. privacy

Personalization has long been framed as a win-win. Customers get relevance. Brands get engagement.

AI turns the volume up.

With enough data, brands can now anticipate needs, predict intent, and tailor experiences in ways that feel almost intuitive. But we face a paradox with Canadian consumers. They want relevance but they are also privacy-conscious. That changes the equation.

The opportunity isn’t pushing personalization further. It’s earning permission to do it at all.

During our conversation, Council members kept returning to the same idea: value exchange. People will share data when they understand how it’s being used and what they get in return.

Trust breaks when:

• Personalization feels invisible or manipulative,
• Data use isn’t transparent, or
• Relevance crosses into intrusion.

AI doesn’t remove the need for consent. It amplifies the consequences when consent is missing.

In practice, that means asking harder questions upfront:

• Is this personalization necessary?
• Is it actionable?
• Would a customer recognize the value if we explained it to them?

Tension two: predictive accuracy vs. human judgement

AI can surface insights at a scale no human team can match. It can spot correlations, anomalies and trends we were never trained to see.

But insight without judgement is just information.

Throughout the discussion, there was strong consensus on this point: AI doesn’t replace human thinking. It shifts where it’s applied.

The most effective teams aren’t handing decisions to algorithms. They’re combining:

• Machine-generated insight,
• Human context,
• Brand understanding, and
• Ethical consideration

Several Council members noted that AI often surfaces things humans were trained to overlook. That’s powerful.

But someone still needs to:

• Question the output,
• Understand the data inputs, and
• Assess bias, relevance and implications.

“Garbage in, garbage out” came up more than once. AI reflects the data, assumptions and biases we bring into it.

That’s not a technical issue. It’s a leadership one.

Tension three: efficiency vs. differentiation

Efficiency used to be a competitive advantage. With AI, it’s becoming table stakes.

When everyone has access to similar tools trained on similar datasets, speed and scale stop being differentiators. Insight risks becoming commoditized.

So where does differentiation come from?

Council members pointed to the same sources repeatedly:

• First-party data,
• Human creativity,
• Brand point of view, and
• Uniquely Canadian context.

AI can help generate options. It can’t decide which ones matter.

One participant described it this way: AI accelerates the work, but humans still choose the direction. The teams that win will be the ones who invest in thinking, not just tooling.

AI changes how we spend our time

One of the most practical takeaways from the conversation was how AI reshapes day-to-day work.

AI doesn’t eliminate roles. It changes the pie chart.

Less time is spent on manual execution. More time shifts to:

• Framing the right questions,
• Feeding better inputs,
• Evaluating outputs, and
• Thinking strategically about audiences and impact.

That shift requires new skills, not fewer people.

Junior analysts, subject matter experts, creatives and strategists all play a role in guiding, validating and refining AI outputs. Insight still starts with a human idea. AI just helps it travel faster.

For tips, best practices and practical insights on leveraging AI effectively amongst your teams, check out the CMA Guide on AI for Marketers.

The question Canadian marketers need to ask now

The most important question that emerged wasn’t about tools or tactics. It was this: How do we build insight-led, human-centred marketing capabilities that reflect Canadian values in an AI-first world?

That means:

• Designing with trust in mind,
• Being transparent about how AI is used,
• Knowing when to lean into automation and when to slow down, and
• Remembering that insight isn’t just discovered, it’s interpreted.

AI is an accelerator, not a decision-maker. It can move us forward faster, but it doesn’t choose the direction.

That responsibility still sits with us.