Session 3 – Learning Scale, AI & AIC²

AI's shift in digital marketing

Traditional digital marketing already collected data, measured behavior, targeted audiences, and automated delivery; AI extends this from understanding the market toward choosing and executing market-facing responses

AI marketing cycle

Observe → Understand → Generate & Choose → Execute & Interact → Feedback

Observe in the AI marketing cycle

Collect market signals such as searches, clicks, reviews, and transactions through foundations such as analytics, CRM, and online listening

Understand in the AI marketing cycle

Use AI to predict, classify, summarize, and explain the market in support of managerial judgment

Generate & Choose in the AI marketing cycle

Generate alternatives and select a response, such as ad copy, a recommendation, audience, bid, or offer

Execute & Interact in the AI marketing cycle

Implement the response in the live market through actions such as serving, displaying, reallocating, updating prices, negotiating, or transacting

Feedback loop in AI marketing

Customer responses become new data for later understanding, choice, and execution

Key shift created by AI

AI increasingly both represents markets and performs within them

Predictive vs. generative AI capability

Predictive AI forecasts, classifies, or scores; generative AI produces, synthesizes, or converses

Representational vs. performative AI role

Representational AI models, interprets, predicts, or explains the market; performative AI acts in or shapes the market

Capability vs. market role

Predictive/generative describes what AI does computationally; representational/performative describes the role its output plays in the market

Can one AI system span multiple capabilities and market roles?

Yes; one system can predict fit, generate a comparison, rank options, and complete a purchase

Data-enabled learning scale

The strategic advantage that arises when increasing market interactions are effectively converted into improvements in future predictions, decisions, offerings, coordination, and actions, thereby creating greater value across the market system

AI data flywheel

More users/interactions → more data → better AI → better product/value → more users/interactions

Does having more data automatically create data-enabled learning scale?

No; the data must be effectively converted into improvements that create greater future value

Why can more data fail to improve learning?

Data can be biased, noisy, low-quality, strategically manipulated, or unrepresentative

Three forms of scale

Cost scale, network scale, and learning scale

Cost scale

More output lowers average cost, creating efficiency and potential cost leadership/differentiation

Network scale

More users and complements raise current value, creating participation value and supporting platform strategy

Learning scale

More interactions improve intelligence and action, allowing offerings and decisions to improve over time and creating adaptive value

Cost vs. network vs. learning scale

Cost scale creates efficiency; network scale creates participation value; learning scale creates adaptive value

Key mechanism of network scale vs. learning scale

Network scale: more users/complements increase current value; Learning scale: more interactions improve future intelligence, decisions, and actions

Potential risks of the three forms of scale

Cost scale: rigidity/diseconomies; Network scale: congestion/concentration/dependency; Learning scale: bias/noisy feedback/opacity/path dependence

When is ecosystem-level value strongest?

When learning scale is combined with complementarity across people, platforms, data, and organizational capabilities

Why does learning scale require a broader marketing framework?

AI learns from market interactions, changes marketing action, and those actions change the conditions that generate the next round of data

Learning-scale market cycle

AI learns from market interaction → learning changes marketing action → action changes future market conditions

AIC² framework

A digital-marketing framework for the AI age organized around Access, Intelligence, Creation × Complementarity, and Co-evolution

Access in AIC²

Who or what becomes visible, reachable, matched, admitted, or excluded—and on what terms

Core question of Access

Who can reach whom or what, and who controls the gate?

Examples of Access mechanisms

Search, feeds, recommendations, AI answers, ad auctions, onboarding, eligibility, targeting, matching, channels, AI agents, devices, APIs, models, and data services

Intelligence in AIC²

Transforms connected signals/data into market representations, predictions, classifications, insights, recommendations, or decision support

Examples of connected signals for Intelligence

Queries, clicks, dwell time, purchases, sharing, prices, and competitor actions

Intelligence-to-action example

Query + device + location + conversion signals → predicted conversion value → bid and budget allocation

Learning-scale caution

Intelligence can optimize the current path while missing better alternatives, so exploration, experiments, and human framing still matter

Creation in AIC²

AI-enabled generation or transformation of market outputs, including content, products, services, experiences, interfaces, messages, design, and market actions

Key distinction about Creation

Creation includes generation AND action; it is not limited to generating content

Examples of Creation

Content/messages, offerings/design, experiences/interfaces, and market or agentic actions such as offers, prices, allocations, negotiation, and purchases

Complementarity in AIC²

AI-enabled resource integration among interdependent actors or capabilities that work together to create value

Digital-ecosystem foundation of Complementarity

Products, services, platforms, and complementors can create more value when used together

Examples of AI-mediated Complementarity

Human + AI collaboration, firm + platform integration, creator ecosystems, employee augmentation, and consumer + AI co-production

What does AI add to complementarity?

AI can lower creation, matching, and coordination costs, but fit and governance still determine whether value is created

C² value engine

Creation × Complementarity; AI-created value depends both on what is created/performed and how well the required actors and capabilities work together

Strong Creation + strong Complementarity

Strongest C² situation; useful AI creation/action is integrated with the data, platforms, channels, workflows, actors, and capabilities needed to make it valuable

Strong Creation + weak Complementarity

A powerful model or output lacks the data, workflow, channel, or governance fit needed to realize its value

Modest Creation + strong Complementarity

A relatively simple AI feature can create substantial value when amplified by strong platform, channel, or workflow fit

Weak Creation + weak Complementarity

A prototype lacks both useful outputs and the users/integration required to create substantial value

Complementarity vs. Co-evolution

Complementarity = working together in a focal activity; Co-evolution = actors and systems changing one another across repeated cycles

Co-evolution in AIC²

Repeated market actions alter actors, data, rules, and the conditions of the next AIC² cycle

Four-step co-evolution cycle

Action → Response → Modification → Next cycle

Example of co-evolution

A recommendation changes customer attention → creators adapt content → the platform retrains or revises ranking → future visibility and data change

Why is co-evolution more than a simple feedback loop?

The responses to an action can change the actors, system, rules, and future market conditions themselves

Agentic co-evolution cycle

More consumers delegate → agent-mediated demand expands → firms reorient product differentiation → catalogs fit agent representation more closely → value of delegation increases

Why can evaluating an AI agent using a fixed market be misleading?

Firms may adapt products and quality to agent-mediated demand, so better choices from today's products do not guarantee better products tomorrow

Fixed catalog vs. adaptive market

Fixed catalog holds products constant and asks whether the agent chooses well; adaptive market recognizes that firms may reposition products, reallocate quality, and respond to agent-mediated demand

Connection in AIC²

Connection is a cross-cutting architecture linking actors throughout Access, Intelligence, Creation, Complementarity, and Co-evolution

Connection in AECCC vs. AIC²

In AECCC, Connect focuses on customer-to-customer conversation; in AIC², Connection broadens to the architecture linking the full market system

Connected market system in AIC²

Consumers, creators, employees, firms, platforms, complementors, institutions, and AI agents interact throughout the market system

AIC² vs. traditional frameworks

Traditional frameworks answer narrower questions; AIC² asks how AI-mediated mechanisms interact and reshape the market

AIC² Diagnostic Canvas

Map Connection → Access → Intelligence → Creation → Complementarity → Co-evolution, while considering governance across the entire cycle

Governance across the AIC² cycle

Ask who benefits, who bears risk, who can contest the system, and who is accountable

Amazon AI shopping ecosystem example

A connected system of customers, sellers, advertisers, devices, logistics, payments, external retailers, and AI

Amazon customer needs for AI shopping

Reduce search/cognitive cost, compare alternatives/tradeoffs, personalize choices and remember context, automate routine tasks, and preserve trust/control/reversibility

Amazon learning-scale loop

Access generates signals → Intelligence turns them into personalized policies/actions → actions change the next signals

Amazon Creation examples

Shopping guidance/comparisons, personalized guides/alerts/carts/purchases, seller insights/listings/growth plans, and advertising creative/prompts

Amazon Complementarity examples

Catalog/review data, sellers/brands/advertisers, apps/devices/payments/logistics, and employees/workflows/governance/disclosure

Main lesson from Amazon's Creation × Complementarity

AI shopping value depends on resource integration across the digital ecosystem—not the model alone

Amazon co-evolution

Consumers delegate → richer signals emerge → actions change → sellers adapt → platform rules evolve → the next cycle creates new Access conditions

Session 3 big picture

AIC² extends traditional digital marketing for the AI age by connecting Access, Intelligence, Creation × Complementarity, and Co-evolution within a learning market system