The AI literacy gap no one in A&M talks about

Imagine a campaign brief landing on a strategist's desk. Before the first stakeholder meeting, an AI tool has already scanned consumer sentiment data, suggested audience segments, generated three creative directions, and flagged which CTA variations performed best in the last quarter. In a media agency, programmatic systems are optimising bids in real time, with machine learning models predicting click-through rates before a single ad goes live. In a martech stack, AI is personalising email journeys for millions of users simultaneously, with no human touching a single send.

This perceived future is already the operating rhythm of many advertising and marketing organisations today. And at the top of the organisational chart, the language has changed accordingly: AI-first, AI-native, AI-powered. WPP has committed over £300 million this year to AI investment, building its open marketing platform and partnering with Google in a $400 million deal. Publicis Groupe has pledged 300 million euros over three years to CoreAI, its internal AI system built on 2.3 billion consumer profiles. Across the holding company landscape, AI is the strategy.

But here is the question that rarely appears in the press releases: who inside these organisations actually understands what that means?

Investments are rising. The skills to match them are not keeping pace.

Global AI spending is projected to exceed $2.52 trillion in 2026. In hiring, the shift is already visible. LinkedIn's Skills on the Rise 2026 report places prompt engineering, workflow automation, LLMOps, and data storytelling among the fastest-growing competencies in India, with prompt engineering now appearing across HR, marketing, sales and consulting, well beyond its technical origins. LinkedIn data also shows that 46% of recruiters globally now rely on skills data to fill roles, while 74% of recruiters in India say finding qualified talent is harder than ever.

The demand within advertising and marketing is specific.Anand V,Chief Information Officer – APAC at Randstad, points out, "Demand is strongest for competencies that sit at the intersection of AI and marketing, such as data analytics, AI-augmented content optimisation, customer journey personalisation, model governance, machine learning literacy and the ability to embed AI into campaign strategy. However, there remains a gap when it comes to advanced technical foundations like NLP or cloud AI proficiency, as well as those who can translate AI outputs into strategic business value."

The gap Anand describes is not simply about whether employees have access to tools. It is about whether they know what to do with the output. And that distinction is increasingly where organisations are falling short.

Many agencies, platforms and brands today position themselves as AI-first. Internally, that is expected to translate into a workforce that can use AI to improve strategy, content, media planning and measurement. But whether formal training exists to support that expectation is a different matter. 

According to a 2025 report from The Adecco Group, 60% of business leaders expect employees to upskill themselves for AI, and yet 34% of organisations admit they have no formal AI policy at work without guidance, safeguards or a plan.

Roopa Badrinath, Founder and Principal Consultant at Turmeric Consulting, shares, "I sometimes wonder if organisations have democratised AI access without democratising AI literacy. Many companies describe themselves as AI-first, but the more important question is: AI-first for whom? For all employees across the organisation, for clients, and for the customers they serve? If the answer is truly 'all of the above,' then AI adoption must move beyond tool rollouts to a more inclusive and deliberate capability-building effort."

Anand echoes this with data-backed concern. 

"While the use of AI tools is becoming commonplace, with a majority of sales and marketing professionals already using them in some capacity, formal and role-specific training remains limited, leaving talent without the deep, practical skills needed to harness AI effectively on the job. This suggests that many companies may be overestimating how AI-ready their talent truly is."

That overestimation has a face at the leadership level too. A Gartner survey found that 65% of CMOs expect AI to dramatically change their role within two years, yet only 32% believe significant changes to their own skills are needed. The report predicts that by 2027, a lack of AI literacy will rank among the top three reasons CMOs are replaced at large enterprises. As Gartner Distinguished VP Analyst Lizzy Foo Kune put it in the report, CMOs cannot treat AI as something the team uses while leadership stays on the sidelines.

The depth of training still lags behind the breadth of adoption

When it comes to who is getting trained and how, the picture is uneven across functions and levels. Most professionals can use AI for drafting, image generation, or pulling research from large documents. Fewer can do the harder things.

Himani Mangtani, Business Head at SW Network, describes the split, "AI exposure today is happening across levels, from junior executives to senior leadership, although the depth of application varies. Within performance marketing and strategy, AI is now being used to build structured frameworks that map target stages, define user actions, generate CTA variations and identify optimisation paths. These functions are currently seeing sharper integration given the measurability of impact."

But that integration is not yet consistent. Mangtani is direct about what is missing at an industry level. She notes, "What is still missing across the industry is structured advanced training around automation workflows, predictive thinking, data interpretation, and responsible AI governance, with most programmes still focused on tool usage rather than system-building."

Bhavya Misra, CHRO at Godrej Capital, sees the same pattern from an HR perspective, noting that teams handling data, performance or customer engagement tend to pick it up faster, but that the real challenge is spreading adoption across the organisation. 

She points to the risk of siloed adoption: if marketing becomes AI-fluent but finance, HR, or operations do not, friction builds. "Adoption must be horizontal, not hierarchical," she says. For marketing specifically, she argues it sits at the intersection of insight, creativity and measurable impact, making it both a natural candidate for AI integration and a critical test case for whether organisations can achieve consistent depth.

The research backs up the importance of that depth. A study led by PhD student Snehal Prabhudesai and Professor Nikola Banovic at the University of Michigan examined how students engage with large language models and whether they can evaluate AI outputs critically. 

Using a framework called PromptAuditor, researchers found that students without structured guidance struggled to identify biases in AI outputs and often defaulted to surface-level interaction. Workshop participants without prior structured AI instruction scored a mean of 66.86 on a 100-point AI literacy scale, compared to 84.78 among classroom participants who had received formal guidance. The study found that targeted instructional support significantly improves the ability to critically evaluate AI outputs. Without it, people use AI without understanding what it is telling them or why. 

The implication for marketing teams is not subtle. When AI is shaping audience insights, creative decisions and campaign strategies, the quality of human judgement applied to that output determines the quality of what reaches consumers.

Roopa Badrinath states, "The quality of any AI output is entirely dependent on the quality of human input — the assumptions, context, and blind spots brought to the keyboard. If employees are not trained to recognise and examine their own biases before constructing prompts, those biases are not mitigated; they are simply automated at scale."

The unspoken expectation: Figure it out yourself

The harder structural question is who is responsible for closing this gap. Formal programmes exist, but they are not yet the norm.

IBM has committed to training 2 million learners in AI by 2026 through its SkillsBuild platform. OpenAI, through its Learning Accelerator, has partnered with India's Ministry of Education, AICTE and six major universities, including IIT Delhi and IIM Ahmedabad, to reach over 100,000 students and faculty with ChatGPT Edu access and structured AI training. Microsoft has expanded its Elevate programme to upskill teachers across schools and higher education in India, working with government agencies. These are broad, sector-agnostic initiatives, and they matter. But they are not filling the gap within advertising and marketing organisations specifically.

Inside agencies, the approach varies. 

Mangtani notes that some of the more substantive capability building is happening not internally but through client relationships. "For certain briefs, especially with large organisations such as Procter & Gamble, we are invited to participate in dedicated training sessions and workshops aligned to their AI frameworks and expectations. This ensures that our teams are not only building internal capability but are also aligned with global best practices and client-specific standards."

BCG research adds a useful benchmark here: only 22% of companies have moved beyond the proof-of-concept phase in AI, and just 4% are creating substantial business value at scale. The gap between pilot and production is almost always a people problem, not a technology one.

Bhavya Misra comments, "Formal programmes help set a shared baseline, bring teams together around governance, and reduce fragmentation. At the same time, AI is moving so fast that curiosity and peer-led experimentation often accelerate adoption. The tricky part comes when exploration starts to feel like an unspoken expectation rather than an encouraged opportunity. Not everyone has the same time, confidence, or exposure, and without clear organisational support, gaps can quietly appear."

That quiet gap is the concern. The unspoken message of 'self-upskill or fall behind' is already being absorbed by employees who sense the stakes but receive no roadmap.

Roopa Badrinath identifies the skills that are hardest to find and hardest to build. "The hardest skills to build and to find are not technical. They are human: critical judgement of AI outputs, data literacy to recognise bias and risk, and the ability to convert AI use into strategic value rather than cosmetic efficiency."

That distinction between cosmetic and substantive AI use is where AI fluency is already beginning to appear in compensation data. Roles that explicitly require AI skills are commanding wage premiums compared to traditional equivalents. Anand V points out, "Organisations are increasingly embedding AI expectations into job descriptions with generative AI and data-driven marketing competencies surging in demand and commanding significant wage premiums compared to traditional skills. On the other hand, structured upskilling programmes are not yet widespread or mature enough to close the gap at scale."

Bhavya Misra puts it most plainly. "Being AI-first is really about changing how work gets done. It cannot just remain a positioning statement; it has to translate into operating discipline. In practice, that means starting with structured literacy at the foundation and then steadily moving into functional integration at the workflow level. It is less about exposing teams to tools and more about embedding AI into planning, optimisation, and decision-making, so it becomes part of the way teams think, not just something they try."

The advertising and marketing industry has always moved at the speed of culture. AI is moving faster. Declaring AI-first intent is now the baseline, not the differentiator. What separates organisations that will genuinely benefit from AI from those that only appear to is whether their people understand it well enough to push back on it, direct it, and build with it. Right now, for most organisations, that workforce does not yet exist at the scale they need.