AI-Managed Consumption Ecosystems and Demand-Shaping Power

1. Introduction

AI-managed consumption ecosystems are markets in which artificial intelligence is used not merely to respond to consumer demand, but to predict, influence, personalize, prioritize, and sometimes actively shape that demand. Examples include AI-driven recommendation engines, digital assistants, algorithmic advertising, personalized pricing, app stores, e-commerce platforms, streaming services, smart-home ecosystems, digital wallets, and AI-powered marketplaces.

The competition-law concern arises when an undertaking with substantial market power can use AI to control the consumer journey from discovery to purchase, thereby influencing which products consumers see, compare, select, and ultimately buy.

Demand shaping can occur through:

• personalized recommendations;
• search-ranking manipulation;
• default settings;
• targeted advertising;
• personalized discounts;
• algorithmic pricing;
• product bundling;
• subscription design;
• loyalty ecosystems;
• self-preferencing;
• interoperability restrictions;
• control over consumer data;
• exclusion of competing suppliers;
• AI-generated product rankings and reviews; and
• prediction of individual consumer willingness to purchase.

The important distinction is between ordinary competition through better recommendations and strategic use of market power to distort competitive conditions.

2. Meaning of Demand-Shaping Power

Traditional competition analysis generally assumes that firms respond to consumer demand.

AI changes this relationship.

An AI-managed ecosystem can potentially:

observe consumer behaviour → predict future preferences → influence exposure → alter consumer choices → collect new behavioural data → improve the prediction system → further influence demand.

This creates a feedback loop.

Traditional market

Consumer preference → demand → firm response

AI-managed ecosystem

Consumer data → prediction → ranking/recommendation → consumer exposure → consumer choice → additional data → improved prediction

The platform may therefore possess power not only over supply, but also over the formation and direction of demand.

3. Competition-Law Framework

The principal competition-law questions are:

A. Relevant market

The relevant market may concern:

• online marketplaces;
• general search;
• digital advertising;
• app distribution;
• mobile operating systems;
• streaming;
• online travel;
• food delivery;
• digital payments;
• cloud ecosystems;
• smart-home platforms; or
• AI recommendation services.

The challenge is that conventional market-share measurements may underestimate power where the undertaking controls consumer attention, data, defaults, interfaces and switching costs.

4. Market Power Through Consumer Interfaces

AI-managed ecosystems may create several layers of market power.

4.1 Data power

Large platforms can collect:

• search history;
• purchases;
• location;
• browsing behaviour;
• viewing habits;
• payment information;
• device information;
• social connections; and
• responses to recommendations.

Data can improve AI models, which can improve recommendations, which can attract more consumers and generate additional data.

This may produce a data-feedback advantage.

4.2 Attention power

An AI system can decide:

• what appears first;
• what is recommended;
• which advertisements are displayed;
• which products receive visibility;
• what consumers are reminded about;
• which alternatives are hidden or deprioritized.

Control over attention can therefore become an important competitive resource.

4.3 Default power

Defaults are particularly powerful because many consumers do not actively change them.

Examples include:

• default search engines;
• default browsers;
• default payment systems;
• default digital assistants;
• default app stores;
• default shopping services.

An AI system can make defaults even more effective by dynamically personalizing the user's experience.

4.4 Ecosystem lock-in

A consumer may simultaneously use:

operating system + app store + cloud + payment service + browser + search engine + digital assistant + smart devices.

Switching one component may require changing several others.

The resulting ecosystem switching cost can protect market power even where individual products face nominal competition.

5. AI as a Demand-Shaping Instrument

5.1 Personalized recommendations

AI can rank products differently for different consumers.

This creates a competition question:

Is personalization improving consumer choice, or is it being used to disadvantage competing suppliers?

A dominant platform could theoretically manipulate recommendation systems to favour:

• its own products;
• affiliated sellers;
• high-margin products;
• products generating more data;
• suppliers agreeing to restrictive terms.

5.2 Self-preferencing

Self-preferencing occurs when a platform gives preferential treatment to its own products or services.

AI can make self-preferencing less visible because discrimination may occur through apparently neutral variables such as:

• predicted conversion;
• relevance;
• engagement;
• consumer satisfaction;
• delivery probability;
• expected retention.

The legal question becomes whether the algorithm's criteria are genuine quality criteria or mechanisms for exclusionary preference.

6. Algorithmic Personalization and Price Discrimination

AI permits increasingly sophisticated forms of individualized pricing.

The system may estimate:

• willingness to pay;
• urgency;
• purchasing history;
• probability of switching;
• price sensitivity;
• consumer lifetime value.

Personalization itself is not necessarily anticompetitive.

However, competition concerns can arise where a dominant firm uses individualized pricing to:

1. exclude rivals;
2. exploit customer lock-in;
3. discriminate against customers who lack alternatives;
4. coordinate prices through common algorithms; or
5. prevent effective price competition.

7. Demand Shaping Through Advertising

AI advertising systems can determine which consumer sees which commercial message.

The platform may therefore control both:

consumer attention and advertiser access to consumers.

This creates potential vertical and horizontal conflicts.

A dominant advertising platform could potentially:

• favour its own advertising inventory;
• disadvantage rival advertising exchanges;
• condition access to consumer data;
• use data obtained from advertisers to compete against them;
• restrict interoperability; or
• manipulate auction mechanisms.

8. Network Effects and AI Feedback Loops

AI ecosystems often exhibit strong network effects.

Direct network effect

More users → more interactions → greater ecosystem value.

Data network effect

More users → more data → better AI → better recommendations → more users.

Commercial network effect

More consumers → more sellers → greater product variety → more consumers.

The combination can produce a self-reinforcing competitive advantage.

A new entrant may therefore face a structural problem:

It needs consumers to obtain data, but it needs data to attract consumers.

9. Six Important Case Laws

The precise concept of "AI-managed consumption ecosystems" is relatively new. Therefore, existing competition cases concerning digital platforms, defaults, self-preferencing, ecosystem control, data and exclusionary conduct provide the principal legal foundations.

Case 1: Google Shopping – European Commission, 2017

Facts

The European Commission found that Google had systematically given prominent placement to its comparison-shopping service while demoting competing comparison-shopping services in its general search results.

Legal principle

The case established the importance of examining how a dominant digital platform controls visibility and access to consumers.

Relevance to AI demand shaping

AI recommendation systems can perform a similar function.

Instead of manipulating traditional search rankings, an AI system could determine:

which products are recommended → which products receive consumer attention → which products obtain sales.

Thus, algorithmic visibility can become a competitive asset.

Competition significance

The case is particularly relevant to:

• self-preferencing;
• ranking;
• platform neutrality;
• consumer access;
• algorithmic discrimination; and
• control over digital demand.

10. Case 2: Google Android – European Commission, 2018

Facts

The European Commission examined Google's conduct concerning Android, including requirements involving Google Search and the Chrome browser and restrictions affecting competing mobile operating systems.

Principle

The case demonstrated how control over an important digital ecosystem can be used to reinforce the position of related services.

Relevance

An AI-managed ecosystem may similarly combine:

operating system + assistant + search + app store + advertising + payments + recommendations.

Control over one layer can influence consumer behaviour at another layer.

Competition concern

The relevant question is whether contractual or technological integration:

• improves the product legitimately; or
• forecloses competing services and strengthens ecosystem dominance.

11. Case 3: United States v. Microsoft Corp. (2001)

Facts

Microsoft was found to have engaged in exclusionary conduct involving its Windows operating-system monopoly and competing browser technology.

Principle

The case illustrates how control over an important platform can be used to influence distribution and consumer access to complementary products.

Relevance to AI ecosystems

The modern equivalent may involve:

operating system → AI assistant → search → recommendation → commerce.

If the platform owner uses control over one layer to restrict competing AI or consumer-facing services, traditional platform-exclusion principles become relevant.

Key lesson

Competition law can examine the strategic use of platform control, rather than looking only at the individual product.

12. Case 4: United States v. Google – Search and Search Advertising

The U.S. Google search litigation has addressed Google's agreements and practices concerning distribution and default search access.

Competition significance

The case illustrates the competitive importance of:

• default status;
• distribution agreements;
• access to users;
• search scale;
• data advantages; and
• network effects.

Relevance to AI demand shaping

AI assistants could become the new consumer gateway.

Instead of a consumer typing:

"best laptop under ₹80,000"

the consumer may ask an AI assistant:

"Which laptop should I buy?"

If the assistant controls the shortlist, the system may become an extremely important demand gatekeeper.

Competition law may consequently need to examine how AI systems determine which commercial options consumers are exposed to.

13. Case 5: Amazon Marketplace – European Commission / Amazon Commitments

Facts

European competition authorities investigated Amazon's use of marketplace seller data and its treatment of competing sellers.

The European Commission accepted commitments concerning the use of non-public seller data and the operation of the Buy Box.

Competition principle

A platform operating a marketplace can possess commercially valuable information about businesses that depend upon the platform.

Relevance to AI

AI substantially increases the potential value of such information.

A marketplace can potentially use seller data to predict:

• which products will succeed;
• which prices will convert;
• which sellers are vulnerable;
• which products should be promoted;
• which products consumers are likely to switch to.

Competition concern

The fundamental issue is:

Can a platform use information generated by dependent businesses to compete against those businesses while simultaneously controlling their access to consumers?

AI can intensify that conflict.

14. Case 6: Epic Games v. Apple

Facts

Epic Games challenged Apple's App Store restrictions, particularly concerning app distribution and payment systems.

The litigation examined Apple's control over app distribution and payment mechanisms within the iOS ecosystem.

Competition significance

The case demonstrates the importance of ecosystem gatekeeping.

Apple controls important elements of:

• app distribution;
• payment mechanisms;
• consumer access;
• platform rules.

Relevance to AI-managed consumption

An AI assistant or AI-powered app marketplace could become an additional gatekeeper.

If consumers increasingly rely upon AI systems to discover applications, products or services, control over the AI interface could affect:

• discoverability;
• consumer acquisition;
• ranking;
• commissions;
• switching;
• payment flows.

Thus, AI recommendation power can supplement traditional platform gatekeeping power.

15. Case 7: FTC v. Facebook / Meta

The U.S. litigation concerning Facebook's acquisitions and conduct examined the competitive significance of network effects and platform ecosystems.

Relevance

Social platforms possess enormous amounts of information concerning:

• consumer behaviour;
• social relationships;
• preferences;
• engagement;
• advertising responses.

AI systems can transform this information into increasingly sophisticated predictions of consumer behaviour.

Competition implication

A platform's competitive advantage may therefore arise not merely from the number of users, but from its ability to convert behavioural data into:

prediction → personalization → engagement → additional data.

This is particularly relevant when analysing barriers to entry.

16. Case 8: Qualcomm

Facts

The Qualcomm litigation in the United States concerned licensing practices and alleged exclusionary conduct involving cellular technology.

Relevance to AI ecosystems

Although Qualcomm is not a consumer-recommendation case, it illustrates a broader principle:

control over an essential technological layer can affect competition in downstream markets.

AI ecosystems may similarly depend on:

• foundation models;
• cloud infrastructure;
• chips;
• APIs;
• operating systems;
• data;
• app stores.

Where one undertaking controls a strategically important layer, downstream competitors may become dependent upon it.

17. Case 9: Intel

Facts

Competition authorities examined Intel's use of rebates and arrangements involving computer manufacturers and distributors.

Principle

Discounts and contractual incentives can become problematic when used by a dominant undertaking to restrict effective access for competitors.

AI relevance

AI-managed consumption ecosystems could employ sophisticated personalized incentives.

For example, a platform might provide:

• individualized discounts;
• preferential commission rates;
• ranking benefits;
• promotional credits;
• loyalty rewards.

The competition analysis would focus on whether these arrangements foreclose competitors, rather than simply on the existence of personalization.

18. Case 10: Booking.com and Online Hotel Distribution

European competition authorities and courts have examined various aspects of online hotel-platform agreements, including parity clauses.

Relevance

Online platforms can influence the competitive relationship between suppliers and consumers through contractual and technological mechanisms.

AI can amplify this by dynamically determining:

• which hotels appear first;
• which prices are highlighted;
• which properties receive recommendations;
• which offers are promoted;
• which customers receive individualized offers.

Thus, algorithmic intermediation can become a form of market power.

19. Common Competition Concerns

A. Self-preferencing

AI may rank the platform's own products above rival products.

Potential theory: abuse of dominance / exclusionary conduct.

B. Discriminatory ranking

Competitors may technically remain on the platform but receive systematically inferior visibility.

This can create de facto exclusion without formal exclusion.

C. Data leveraging

The platform may combine:

• consumer data;
• seller data;
• advertising data;
• payment data;
• search data.

The resulting data advantage can make market entry increasingly difficult.

D. Tying and bundling

AI assistants may be bundled with:

• operating systems;
• search;
• browsers;
• cloud;
• payments;
• shopping;
• advertising.

The competition issue arises where control of one product is used to extend dominance into another market.

E. Exclusive distribution

An AI ecosystem may restrict suppliers from simultaneously using rival ecosystems.

This can increase switching costs and reduce multi-homing.

F. Algorithmic exclusion

An AI model might systematically reduce the visibility of rival suppliers.

Unlike an explicit contractual prohibition, algorithmic exclusion may be difficult to detect because the platform can characterize the outcome as the product of neutral optimization.

20. The "Black Box" Problem

One of the most important issues is explainability.

Suppose a platform says:

"Our AI ranks products according to relevance."

Competition authorities may need to determine:

• What does "relevance" mean?
• Which variables influence the ranking?
• Does commission influence ranking?
• Does ownership influence ranking?
• Does advertising expenditure influence ranking?
• Are competing suppliers penalized?
• Does the model learn from previous rankings?
• Were competing products systematically exposed to less traffic?

The difficulty is that AI systems can continuously change their behaviour.

Therefore, conventional evidence may be insufficient.

21. Algorithmic Feedback and Entrenchment

An especially significant problem is feedback-based exclusion.

Consider:

Platform promotes Product A

↓

Product A receives more clicks

↓

AI observes higher engagement

↓

AI predicts Product A is more relevant

↓

Product A receives even greater ranking

↓

Competitor B receives fewer clicks

↓

AI interprets B's low engagement as evidence of low consumer interest.

This creates a self-fulfilling algorithmic preference.

The platform may then argue that the algorithm is merely responding to consumer behaviour, even though the initial allocation of visibility helped create that behaviour.

22. Consumer Choice Architecture

Demand-shaping power also operates through interface design.

Examples include:

• pre-selected options;
• default subscriptions;
• auto-renewal;
• recommended bundles;
• "frequently bought together";
• personalized urgency messages;
• countdowns;
• AI-generated comparisons;
• default payment methods.

Competition law may intersect with consumer-protection law where these mechanisms affect both consumer autonomy and competitive access.

23. AI and Switching Costs

AI systems can create highly personalized ecosystems.

A consumer may accumulate:

• preference histories;
• playlists;
• purchase profiles;
• smart-home configurations;
• recommendation histories;
• loyalty benefits;
• personalized AI instructions.

The longer the consumer remains within the ecosystem, the more valuable the personalization becomes.

This may produce:

personalization → switching cost → retention → more data → stronger personalization.

Such dynamics can increase barriers to entry.

24. Essential Facility-Type Concerns

In certain circumstances, an AI ecosystem may become an important gateway to consumers.

Potential examples include:

• dominant app stores;
• dominant search engines;
• major AI assistants;
• dominant digital advertising exchanges;
• major online marketplaces.

A refusal to provide access is not automatically unlawful.

However, competition analysis may consider:

1. whether the facility is genuinely indispensable;
2. whether effective alternatives exist;
3. whether refusal eliminates effective competition;
4. whether access can technically be provided;
5. whether legitimate business justification exists.

25. Merger-Control Implications

AI-managed consumption ecosystems also create significant merger concerns.

A transaction involving:

AI assistant + marketplace

could create incentives to steer consumers toward affiliated products.

Similarly:

AI model + advertising platform

could permit integration of prediction and advertising.

Or:

AI recommendation engine + payment system

could enable control over the entire consumer journey.

Traditional turnover thresholds may not always capture the strategic importance of emerging AI firms, making transaction value and innovation theories of harm potentially relevant.

26. Relevant Theories of Harm

Competition authorities could investigate:

1. Foreclosure

Rivals lose meaningful access to consumers.

2. Raising rivals' costs

Competitors must pay more for:

• advertising;
• distribution;
• data;
• platform access;
• API access.

3. Self-preferencing

The platform privileges its own services.

4. Tying

Users are effectively required to adopt an associated service.

5. Exclusivity

Suppliers or consumers are discouraged from using rival platforms.

6. Exploitative data practices

The platform obtains excessive informational advantages from ecosystem participants.

7. Coordinated effects

Common AI systems may facilitate parallel or coordinated conduct among firms.

27. Difference Between Legitimate and Problematic Demand Shaping

The existence of AI alone does not establish an infringement.

The relevant issue is the relationship between market power, conduct, effects, and legitimate justification.

28. Evidence Required by Competition Authorities

AI-related investigations may require analysis of:

• source-code documentation;
• model-development records;
• training-data policies;
• ranking variables;
• recommendation logs;
• A/B testing;
• consumer-exposure data;
• click-through rates;
• conversion rates;
• internal communications;
• pricing experiments;
• seller dashboards;
• API access records;
• model-output histories;
• changes to algorithms over time.

Particularly important is counterfactual analysis:

What would consumers have seen if the dominant platform had not applied the disputed algorithmic mechanism?

29. Remedies

Possible competition remedies include:

Structural remedies

• divestiture;
• separation of business units;
• restrictions on acquisitions.

Behavioural remedies

• non-discrimination obligations;
• ranking transparency;
• interoperability;
• data portability;
• access obligations;
• restrictions on self-preferencing;
• prohibition of discriminatory API practices.

Algorithmic remedies

• independent audits;
• algorithmic monitoring;
• logging requirements;
• explainability requirements;
• testing for discriminatory ranking;
• preservation of historical model versions.

Consumer-facing remedies

• choice screens;
• alternative defaults;
• easy switching;
• data portability;
• interoperability.

30. Indian Competition-Law Relevance

For India, the principal framework is the Competition Act, 2002, particularly:

• Section 3 – anti-competitive agreements;
• Section 4 – abuse of dominant position;
• Sections 5 and 6 – combinations;
• Section 19 – inquiry by the Competition Commission of India;
• Section 26 – investigation procedure;
• Section 27 – orders against abusive conduct.

AI-managed consumption ecosystems could potentially implicate Section 4 where a dominant digital platform uses:

• discriminatory conditions;
• discriminatory access;
• tying/bundling;
• denial of market access;
• leveraging;
• exclusionary pricing;
• preferential treatment;

to distort competition.

The CCI's digital-market jurisprudence, including matters concerning Google, e-commerce platforms and digital ecosystems, provides an important foundation for applying established competition principles to AI-mediated consumer markets.

31. Doctrinal Synthesis of the Case Laws

The cases collectively demonstrate several recurring principles:

32. Emerging Legal Doctrine

The central competition-law development is a shift from examining only:

"Who supplies the product?"

toward examining:

"Who controls the consumer's path to the product?"

AI makes this distinction particularly important.

A company may not manufacture a product, own the seller, or set the final retail price, yet it may control:

discovery → ranking → recommendation → advertising → comparison → purchase → payment → post-purchase engagement.

That creates a new dimension of intermediation power.

33. Conclusion

AI-managed consumption ecosystems can transform market power from control over production and supply into control over consumer attention, prediction and choice architecture.

The most significant competition-law risks arise where a powerful ecosystem combines:

data + AI prediction + consumer interface + defaults + recommendation + distribution + payment + ecosystem lock-in.

The key legal question is not whether an AI system influences consumer demand—successful businesses routinely do so through innovation and marketing. The more difficult question is whether a firm with substantial market power uses AI-driven demand shaping to foreclose rivals, discriminate against competing suppliers, extend dominance into adjacent markets, exploit ecosystem dependence, or otherwise distort the competitive process.

The existing jurisprudence on Google Shopping, Google Android, Microsoft, Amazon, Apple, Facebook/Meta, Qualcomm, Intel and digital distribution platforms provides the doctrinal foundation, while AI introduces new problems concerning algorithmic opacity, feedback loops, personalized ranking, data advantages and automated control of consumer choice.