Expert Insights

Q1: What are the real risks marketing teams are already facing with AI adoption in their day-to-day work?

The pressure to deploy AI quickly is real and in many organizations, governance has not kept pace. The most immediate risks are not hypothetical. They are showing up in daily workflows right now.

The most common gaps include:

• Personal and sensitive data is being fed into AI models without compliance checks, creating liability exposure for organizations and risk for consumers
• Teams using personal AI accounts rather than business-licensed tools, inadvertently train models on private or proprietary information
• Content going live without any human review process, creating quality and brand consistency issues downstream
• A wide and largely unacknowledged gap in AI literacy across teams — some employees are highly capable, others are entirely unfamiliar, and most organizations are not addressing that range with any structured training

There is also a broader operational problem at play. The urgency to adopt AI has led many organizations to skip the foundational steps that would normally accompany any significant technology investment. As one speaker put it, if you handed someone Salesforce and told them to go figure it out, you would expect chaos. That is effectively what is happening with AI right now in a lot of companies. The result is not a lack of usage — it is usage without strategy, governance, or measurable return.

Harvard research has shown that many organizations investing heavily in AI are not seeing the returns they expected. That is not an AI problem. It is an AI operations problem.

Q2: What causes AI to lose a brand’s voice when creating content at scale and how can teams avoid it?

Brand voice degradation in AI-generated content is more common than most teams realize and it often happens gradually, not all at once.

Two root causes stand out:

1. The brand voice was never clearly defined to begin with. Many organizations assume their brand voice is documented and understood. Often, it is not. Without a clear, consistent definition, one that captures not just tone but the audience it is speaking to and the value it is meant to deliver, there is nothing solid to train an AI system against.

2. AI does not hold context over time. Even when a brand voice is properly defined, language models degrade over extended interactions. Without continuous reinforcement, re-introducing context, correcting drift, maintaining prompts, the output gradually reverts toward a generic mean. The brand ends up sounding like a caricature of itself rather than the real thing.

The more useful way to think about AI in a content context is as an aggregator, not a creator. It produces the average of what it has been fed. That can be genuinely useful for high-volume, informational content. But it will not produce something novel, emotionally resonant, or award-winning on its own. The brands that use it well treat it as a starting engine, not a finishing one.

A practical rule of thumb: for informational content, AI can get you most of the way there. For anything externally facing that requires emotional resonance, storytelling, or a specific human perspective — put a human on the last mile. An 80/20 split is a reasonable operating principle for most teams.

Q3: Where should marketers draw the line between automation and human input?

The clearest signal for where automation should stop is emotion.

When a customer is in a heightened emotional state, frustrated, anxious, confused, or genuinely excited, they need a human to absorb and respond to that energy. A chatbot cannot do this. It does not just fail to help; it actively makes the experience worse. Anger escalates. Excitement deflates. The customer is left feeling that no one cares enough to show up for them.

A simple litmus test for where automation belongs:

• Use automation where the interaction is transactional, the customer wants speed and convenience, and there is no significant emotional stakes
• Bring in humans where the situation involves frustration, urgency, vulnerability, or any moment where a customer needs to feel genuinely heard

The mistake most brands make is underestimating how many customer interactions carry emotional weight. It shows up in places teams do not anticipate. Not just in complaints, but in routine service interactions where a customer’s underlying frustration with being automated away has already been building.

Consumers today are also far more perceptive than brands give them credit for. They can identify AI-generated content by its tone and sentence structure. They know when they are talking to a bot. And in a cultural moment where many people already feel that large organizations do not care about them, the experience of being handed off to an automated system at a critical moment confirms exactly that.

Q4: How is AI affecting consumer trust in brands and what should marketers do to maintain transparency and credibility?

The trust challenge AI creates for brands is structural, not just tactical.

As more organizations race to showcase AI capabilities, they are simultaneously creating what one speaker described as tech parity — a landscape where everyone is using the same tools to produce the same outputs, and differentiation disappears. When that happens, the moat shifts. The competitive advantage stops being the technology and starts being the human element that brands have been quietly removing.

The core tension: Consumers build trust relationally. Brands scale operationally. When brands optimize for operational scale at the expense of human connection, loyalty does not follow. You cannot automate your way to genuine customer relationships and still expect the emotional loyalty that comes from human ones.

On the data transparency front, the issue is not just privacy — it is the feeling of being tricked. Consumers are not necessarily opposed to their data being used. What they object to is not knowing how it is being used, feeling like they have no control over it, and receiving nothing meaningful in return. The brands that earn confidence are the ones that are clear about what they are collecting, why they are collecting it, and what value it creates for the customer.

Spotify’s Wrapped is a useful example of this done right — data is collected, fed back to the user in a way that feels personal and celebratory, and shared willingly. That is transparency with a value exchange built in.

What brands should do:

• Be explicit about where and how AI is being used in customer interactions
• Build feedback loops between AI outputs and human review before content or communications go live
• Invest in genuine human touchpoints — events, direct conversations, moments of real connection — rather than treating them as inefficiencies to be automated away
• Treat colloquialisms, cultural context, and genuine personality as features, not risks. The penalty for sounding too human is far smaller than the penalty for sounding like everyone else

The brands that will build lasting trust are not the ones deploying the most AI. They are the ones that use it thoughtfully and remain recognizably, irreducibly human where it matters most.