Marketing
The relevant question isn ' t who knows how to prompt a large language model. It ' s who can ask good questions before trusting an output, who can hold ambiguity without stalling the team, and who is willing to take ownership when an AI-influenced decision doesn ' t land.“ The goal isn ' t better AI. It ' s better human decisions with AI.”
MARKETING AFRICA
Marketing
Barbra Kingori
AI Doesn’ t Make Marketing Decisions. Marketers Do!
Dr. Elena Alikhachkina Research
New research from Dr. Elena Alikhachkina reveals the missing layer in marketing ' s AI transformation and it isn ' t the algorithm.
Every marketing leader today has access to roughly the same AI tools. Far fewer have figured out how to make good decisions with them.
Most of the industry conversation about AI in marketing still centres on the technology, which platform, which model, which plug-in to adopt next. But new research presented at Oxford ' s Human-AI( HAI) workshop suggests the real gap sits somewhere else entirely: not in the algorithm, but in how humans govern the decisions AI increasingly shapes.
The research comes from Dr. Elena Alikhachkina, a four-time Chief Data & AI Officer, board director, and founder of InfoFluence™. Over three years, across more than 40 organizations spanning healthcare, consumer goods, manufacturing, financial services, and technology, she set out to answer a simple question: how do people make decisions in increasingly data, technology, and AI-driven environments? Her conclusion has direct implications for how marketing teams should think about their own AI transition.“ The missing layer of AI governance isn ' t the algorithm. It ' s us.”
AI Doesn ' t Learn. Your Organization Does.
The starting point of Alikhachkina ' s research is a myth she found repeated across nearly every organization she studied: that AI improves itself. It doesn ' t. What actually improves is the system around it, the data it ' s fed, the decisions people make with its output, and the corrections humans apply afterward.
She maps this as a four-stage loop. Infor-,
mation is the foundation: clean data, business context, policies, and feedbackbecause AI is only as good as the information it receives. AI is the engine, it identifies patterns, generates predictions, and recommends actions, but as she puts it plainly:“ AI recognizes patterns. It does not understand your business.”
Decisions and Actions is where people use AI ' s insights to change business processes and create outcomes. AI suggests; humans decide. And Human Learning is the multiplier: validating results, correcting mistakes, improving prompts and data, adjusting processes. Organizations learn, and AI reflects those improvements.
For a marketing team, this loop plays out constantly and mostly invisibly. Clean customer data and brand context go in. The AI model surfaces a pattern: an audience segment, a content recommendation, a predicted churn risk. A marketer decides what to do with it. And then, critically, the team has to notice what worked, correct what didn ' t, and feed that learning back into the system. Skip that last step, and the loop breaks, the tool stops improving no matter how sophisticated it is.
The Governance Gap in Marketing AI
Ask most marketing organizations what their AI governance covers, and you ' ll hear a familiar list: model governance, data quality, privacy and security, ethics and bias, compliance, monitoring, explainability. All necessary. All, according to Alikhachkina ' s research, incomplete.“ Traditional gover-nance protects AI systems. It rarely prepares people to make better AI-enabled decisions.”
Her research identifies what ' s missing from almost every governance framework she reviewed: how people ask better questions, how they interpret AI outputs, how they calibrate trust, how they communicate uncertainty, how they challenge assumptions, how they take accountability, and how they turn AI insight into action. She calls this missing layer Human Decision Governance.
Translated into marketing terms, this is the difference between a checklist and a capability. A checklist confirms the AI-suggested audience segment doesn ' t violate a privacy policy. A capability asks whether the segment actually makes strategic sense, what data trained it, and what it might be missing. A checklist confirms AI-generated creative passed a brand-safety filter. A capability asks whether a human actually reviewed it before it is shipped under the brand ' s name.
Digital Decision Fluency: The Capability Marketing Has Been Missing
From her research, Alikhachkina developed
The relevant question isn ' t who knows how to prompt a large language model. It ' s who can ask good questions before trusting an output, who can hold ambiguity without stalling the team, and who is willing to take ownership when an AI-influenced decision doesn ' t land.“ The goal isn ' t better AI. It ' s better human decisions with AI.”
72 MAL73 / 26 ISSUE