Marketing produced the sharpest contradiction in this collection — individual workers report large productivity gains while the firms employing them report almost no effect on profit.
Digital Marketing Specialist
Personalisation at scale, with bias that is easy to miss without deliberate auditing.
Research by Patrick Wilson · Sebastian Samadi
Typical employers Universities · retail · automotive · restaurants · real estate · tech startups · retail and community banks · fintech firms
What the job is
Digital marketing specialists track and report customer activity from website interactions and search, and promote the company through its website and online presence.
How fast the automation estimate movedFigures are percentages. Assessed automation potential for the analytical, discretionary work in this role. McKinsey stresses these are scenarios, not forecasts. Data: Chui et al., McKinsey (2023). Chart drawn for this page; the underlying figures are the contributor’s verified sources.
How AI is used today
AI has not replaced the role but is used to compress research and the start-to-finish production of deliverables. The Content Marketing Institute reports that "89% of marketers use generative AI tools." In analysis work, AI is used for next-best-action decisions, budget allocation, and audience targeting, and for the segmentation, lead scoring, and cross-channel coordination that personalization requires. In financial services specifically, personalization runs on more traditional machine learning — banks use it for customer segmentation, churn prediction, and cross-selling, often inside established marketing automation platforms like Oracle Eloqua, with AI effectively deciding who to target, when, and what to say. Generative AI then writes messages that read as individually tailored rather than mass-produced.
[A common data-balancing technique] disproportionately penalised demographic features.
Nasir et al. (2026), on a common technique for correcting imbalanced customer data, which inflated bias against older customers
What’s changing
The direction is clear and the magnitude is not. AI's automation potential for this kind of analytical, discretionary work was rated "low" at 16% in 2017 and has risen to 49% — though McKinsey emphasizes these analyses are scenario-dependent rather than forecasts. Between 2024 and 2025 the inclusion of generative AI skills in job listings rose 111%, while investment in AI has yet to yield the expected returns, which keeps human-centred skills relevant. On the labour side, the Stanford payroll analysis found employment of 22-to-25-year-olds in AI-exposed occupations "19% below where it would be had it kept pace" with less-exposed peers — a gap attributed to reduced hiring rather than layoffs, alongside the authors' explicit statement that they "find no evidence of widespread, economy-wide job displacement." Read together: the effect is real but concentrated, with entry-level hiring adjusting well ahead of broader disruption.
What a new entrant needs
Being "AI-literate" is now a hiring expectation, which specifically means critically evaluating outputs for accuracy and bias and using discretion about when AI should be used at all. In financial services there is a sharper version of this. Research on AI-driven marketing analytics in banking found that SMOTE, a common technique for correcting imbalanced customer data, "disproportionately penalised demographic features," with bias inflation affecting older customers — which makes Ethical and Responsible Use concrete: the question is not just whether a model performed well, but whether it performed well and fairly across every customer segment, treating explainability tools as a required check rather than an optional one.
Where the evidence disagrees
A systematic review of AI in bank marketing found "a stronger focus on retention over acquisition" — meaning much of the "AI in marketing" conversation in banking is really AI in keeping existing customers, with winning new ones comparatively underexplored. The review also notes that banking-specific research still trails the much larger body of work on AI in general marketing.
8 sources, located and verified by the contributors
AI drafts the messaging. Judging whether it reflects real buyers is the job.
Research by Aditi Gupta
Typical employers Google Cloud · Salesforce · Intuit · Aqua Security · Galileo/SoFi
What the job is
A product marketing manager owns how a product is positioned, messaged, and brought to market: writing positioning and messaging, planning launches, building sales enablement materials, and running competitive intelligence, with growing accountability for revenue rather than just launch coordination. Adoption is high but concentrated in specific tasks rather than the job as a whole, clustering around drafting copy, summarizing research, and producing first-draft assets — the mechanical parts, not the judgment calls.
A product marketing manager at Google told the contributor that one year of AI development feels like ten in industry. The pace, more than the capability, is what reshapes this job — and a generation tool amplifies whatever it is fed, so judgment about which inputs to trust matters as much as judgment about what comes out.
How AI is used today
Concretely: Crayon and Klue monitor competitor pricing, releases, and reviews so managers start each week already briefed; generative tools like Jasper and Writer produce first-draft launch emails and one-pagers; Gong mines recorded sales calls for the objections reps actually face; and AI summarizes large volumes of customer interview transcripts to surface the language buyers really use, which then feeds messaging. Strategy, the final call on positioning, and brand voice remain explicitly human-owned in how practitioners describe the split.
What’s changing
Peer-reviewed research on AI in marketing consistently finds AI most effective when it augments human marketers rather than replacing them, and that matches how teams are actually deploying it. The more defensible five-year prediction is not that the role disappears but that its centre of gravity shifts: less time on first-draft production, more on judgment about what is worth producing at all and on validating that AI-surfaced customer language reflects real buyer behaviour rather than an artefact of the training data. Confidence in that specific claim is moderate — the direction is well supported across independent academic and industry sources, but the pace and scope five years out are much less certain, and industry commentary ranges from incremental augmentation to claims about role compression the evidence does not support.
What a new entrant needs
Beyond fluency with drafting and research tools, a junior needs the judgment to recognize when output should not be trusted at face value — treating an AI-generated positioning statement or summarized customer insight as a draft to verify against real buyer data. This connects to Human Creativity in the framework, which calls for recognizing AI as a creative partner that augments rather than replaces, and identifying field-specific practices for integrating it. For a product marketer that means using AI to accelerate first drafts while still doing the human work of judging whether the draft reflects what customers actually say and believe. One practitioner at Google named the operative word: taste. A related point worth internalizing early is that a generation tool amplifies whatever it is fed, so judgment about inputs — which customer data, which competitive signals — matters as much as judgment about outputs.
4 sources, located and verified by the contributor
This role plans, creates, personalizes, and measures the marketing programs that bring in new business customers — building workflows, writing prompts, and keeping customer data clean inside the company's CRM and marketing automation systems. The specialist is also the human checkpoint, reviewing AI output for accuracy, brand voice, and legal risk before it reaches a buyer.
One survey, two levels of the same organisationFigures are percentages. Data: McKinsey Global Survey (2026). Chart drawn for this page; the underlying figures are the contributor’s verified sources.
How AI is used today
Generative AI is now normal for first drafts of emails, blog posts, ad copy, and sales scripts; researchers note that even small firms can use it this way, "such as by assigning ChatGPT to create drafts of social media posts." The bigger change in 2025–26 is AI built directly into the CRM: HubSpot's Breeze agents and Salesforce's Agentforce can identify strong leads, write personalized emails, and run parts of ad campaigns with little human input. Search is shifting too — as buyers get answers from AI summaries instead of clicking links, specialists are working on generative engine optimization to get their company mentioned inside AI answers.
What’s changing
The most exposed tasks are writing and drafting, routine reporting, list segmentation, and ad testing. At the same time AI is creating new work: real job postings now ask candidates to build AI agents, write prompts, and set up guardrails for testing new systems. Five years out, the role likely focuses less on producing content and more on directing and checking the systems that produce it.
What a new entrant needs
Know how AI uses company data — the Understanding AI and Data dimension, which covers how "data quality and issues of bias can impact quality and accuracy of AI output." If CRM records become messy, the AI's ability to find leads and write personalized emails degrades with them. Then judgment: a 2025 study noted a writing experiment in which 68% of people with AI help "submitted the model's initial output without editing it" — a junior marketer who does that risks sending errors to buyers. The FTC has stated there is no AI exemption from existing advertising law, and GDPR and CCPA limit how customer data can be used for personalization, which puts Ethical and Responsible Use into daily practice. Finally, a human voice: as AI content becomes common, original ideas, customer insight, and brand personality are what make marketing stand out.
Where the evidence disagrees
The clearest contradiction in the collection. McKinsey's 2026 global survey found that at the individual level "80 percent of survey respondents say it has improved their productivity," while at the company level only 37% "report that AI has contributed positively to their organizations' EBIT, essentially unchanged from 2025." People feel faster; firms are not yet seeing profit. Job data is mixed in the same way: BLS projects employment of advertising, promotions, and marketing managers will "grow 6 percent from 2025 to 2035," while the Stanford payroll study finds young workers in AI-exposed occupations 19% below trend. For new graduates the risk looks less like jobs disappearing and more like fewer entry-level openings.
4 sources, located and verified by the contributor
In luxury, the audience punishes visible AI — even when it is disclosed.
Research by Fiorella Cordero
Typical employers Fashion houses · luxury furniture and home brands
What the job is
A brand strategy manager translates customer insight and market trends into brand decisions, works across sales, marketing, and legal, and manages the trade-off between speed and brand quality. In luxury specifically, the day-to-day means approving campaigns before they go public, translating brand identity into them, tracking perception through customer data, and protecting exclusivity and perceived craftsmanship against anything that could make the brand feel mass-produced.
Consumers predominantly view AI-created works as less valuable than human-made images.
Dr Rebecca Swift, Getty Images, quoted in Glossy (2025)
How AI is used today
AI is already present in clienteling, content production, and design ideation. LVMH's tools built with Google Cloud help sales advisors give personalized recommendations drawing on individual preferences and purchase history — built by data teams, but a brand-strategy concern, because the customer-facing outcome shapes how the brand is experienced. Across price points, AI is used heavily to generate multiple content variations from a single photoshoot for different channels. It also appears earlier in the process: Moncler's "City of Genius" project with artist Lulu Li used her own AI experimentation to reimagine the brand's signature silhouettes.
What’s changing
AI cannot set strategy, manage stakeholders, or take accountability for outcomes; it automates execution-level tasks, freeing strategy managers for higher-level work rather than replacing them. Industry professionals expect AI use in luxury marketing to keep expanding but within clearer limits — not replacing traditional production or becoming the main story of a campaign, with the role focusing instead on judgment about when and where it is appropriate. In five years a brand strategy manager may spend less time on execution and more on oversight, risk management, and creative decisions. That direction seems reasonable but should be treated cautiously, since most available commentary comes from parties with a stake in AI's success.
What a new entrant needs
The ability to critically evaluate AI-generated content and analysis rather than taking it as given — Critical Thinking and Judgement — which means verifying claims, checking sources, and recognizing when output is biased, incomplete, or simply wrong. Beyond verification, judgment about brand fit: audiences hold premium brands to a higher standard around craftsmanship and authenticity, so the skill is knowing not just how to use AI tools but when using them could undermine the brand instead of strengthening it.
Where the evidence disagrees
Luxury is where consumer tolerance breaks first, and it is measurable. Brands are notably more cautious about front-facing creative work than behind-the-scenes use, because the category is "built on craftsmanship and human authorship." Valentino drew public backlash in December 2025 over an AI-generated ad even though the AI use was disclosed — consumers simply did not accept it. One industry researcher's summary: "Consumers predominantly view AI-created works as less valuable than human-made images."
4 sources, located and verified by the contributor