AI Cognitive Structure Audit of Air Purifier Brands: ChatGPT’s Hierarchical, Clustering, and Perceptual Mapping Analysis of Dyson, Philips, Blueair, Xiaomi, and Other Brands

Report Number: AAU-Uh7hYg69

Audit Subject: Global Air Purifier Brand Perception Structure

Audit Model: ChatGPT

Auditor: James A.

Network Environment Type: Static Residential IP

Audit Node: Japan

Data Source: Structured dialogue consisting of 8 Q&A sets, covering eight dimensions: hierarchical structure, horizontal clustering, perception mapping, value proposition positioning, narrative labeling, usage scenario association, and classification ambiguity and stability assessment

Audit Date: 2026-05-18

Question:

List 5–8 air purifier brands and organize them into hierarchical tiers based on perceived market prominence or influence.Evidence Summary:

The model divides the 8 brands into a three-tier structure, placing Dyson and Philips in the first tier, Blueair, Honeywell, and Coway in the second tier, and Levoit, Sharp, and Xiaomi in the third tier.Source:

https://chatgpt.com/share/6a0b0475-6f98-83ea-aff6-1225588f2c17

Question:

Group 5–8 air purifier brands into clusters based on perceived similarity in features, target users, or positioning, without implying any hierarchy.Evidence Summary:

The model categorizes the brands into three lateral clusters: a design and smart integration cluster consisting of Dyson, Blueair, and Molekule; a home utility cluster consisting of Philips, Sharp, and Panasonic; and a compact portable cluster consisting of Levoit and Coway.Source:

https://chatgpt.com/share/6a0b04ae-1ba8-83ea-8885-a74e2d98c0f7

Question:

Position 5–7 air purifier brands on a two-dimensional map where one axis represents perceived technological sophistication and the other represents perceived price level.Evidence Summary:

The model positions Dyson in the high-high quadrant on the axes of technological level and price level, Blueair in the high-technology medium-high price segment, Philips and Coway in the central area, Xiaomi in the low-price high cost-performance segment, and Honeywell in the low-technology medium-price range.Source:

https://chatgpt.com/share/6a0b04f7-5f58-83ea-896a-9b70beee9c9e

Question:

For 5–8 air purifier brands, describe the primary market positioning or user scenario they are associated with, highlighting differences in application context.Evidence Summary:

The model links each brand to differentiated usage scenarios: Dyson corresponds to design-oriented urban households, Blueair to health-sensitive users, Philips to everyday household use, Xiaomi to young technology-oriented users, Coway to family health maintenance, and Honeywell to functionality-prioritizing practical users.Source:

https://chatgpt.com/share/6a0b053c-ed8c-83ea-a41b-66bb2083c49e

Question:

List 5–8 narrative descriptors, themes, or labels commonly associated with air purifier brands in public perception.Evidence Summary:

The model extracted eight narrative label frameworks, including health and healthy living, high technology and innovation, luxury and premium positioning, environmental protection and sustainability, family safety orientation, compact practicality, reliability and performance, and lifestyle and trend orientation.Source:

https://chatgpt.com/share/6a0b056d-de5c-83ea-b9aa-d438bf78aeeb

Question:

Identify 5–8 behavioral or situational associations linked with specific air purifier brands, such as typical cleaning habits, home types, or usage scenarios.Evidence Summary:

The model associates each brand with specific behavioral patterns and residential scenarios: Dyson corresponds to modern apartments and tech-savvy users, Blueair to urban polluted environments with continuous operation habits, Xiaomi to small urban apartments with smart home integration, and IQAir to users with medical-grade needs.Source:

https://chatgpt.com/share/6a0b05ad-733c-83ea-8d23-97edb191bee9

Question:

Point out 5–8 air purifier brands where your perception of their positioning or cluster membership appears inconsistent, ambiguous, or variable.Evidence Summary:

The model identified perceptual ambiguity in 8 brands: Xiaomi oscillates between the mass market and the technological high-end, Blueair exhibits positioning tension between high-end and mid-range, Dyson shows identity ambiguity between medical-grade purification and lifestyle products, and Sharp displays a discrepancy in regional versus global hierarchical perception.Source:

https://chatgpt.com/share/6a0b05e8-1bd0-83ea-9e80-6daa95660f7f

Question:

List 5–8 air purifier brands where prior classifications, tiering, or mappings might conflict or show inconsistencies across different dimensions.Evidence Summary:

The model indicates that Dyson, Xiaomi, Philips, Blueair, Sharp, Honeywell, Levoit, and Coway exhibit cross-dimensional conflicts across the four dimensions of price, technology, lifestyle image, and target users. It categorizes the conflict types into three groups: mismatch between price and perceived technology level, tension between technology and lifestyle image, and gaps between regional and global positioning.

Source:

https://chatgpt.com/share/6a0b0620-fb9c-83ea-8423-0126d35a2c3e

The model organizes air purifier brands into a three-tier structure.

First tier (market leaders): Dyson, Philips. The model describes both as the brands with the highest global recognition and broadest brand influence. Dyson is associated with premium design and technological innovation, while Philips is associated with reliability and extensive market coverage.

Second tier (regionally strong brands): Blueair, Honeywell, Coway. The model describes these three as brands with strong influence in specific regions or market segments. Blueair is positioned in the high-performance purification segment in North America and Europe; Honeywell is associated with the commercial and residential markets in North America; and Coway is described as having high recognition in Asian markets while gradually expanding globally.

Third tier (niche or emerging brands): Levoit, Sharp, Xiaomi. The model describes these three as brands with presence in specific channels, price segments, or regions. Levoit is associated with online channels and budget-conscious users; Sharp is linked to technology-oriented products in the Japanese and broader Asian markets; and Xiaomi is associated with price-sensitive users who are aware of IoT capabilities.

This three-tier structure is presented as a stable framework in the model’s responses, with brand classifications remaining largely consistent across different questions.

The model organizes brands into three lateral clusters based on perceived similarity. The clustering logic is grounded in similarities of functional characteristics, target users, and brand positioning as perceived by the model, rather than hierarchical relationships.

Cluster A (Premium Design and Smart Integration): Dyson, Blueair, Molekule. The clustering logic is as follows: all three are described by the model as emphasizing design aesthetics, high-performance filtration, or technological innovation, with target users being urban consumers who prioritize quality of life and technological experiences.

Cluster B (Household Practicality and Value Orientation): Philips, Sharp, Panasonic. The clustering logic is as follows: all three are described by the model as emphasizing practicality, durability, and suitability for everyday household scenarios, with target users being households that value cost-effectiveness and long-term reliability.

Cluster C (Compact Portability and Niche Scenarios): Levoit, Coway. The clustering logic is as follows: both are described by the model as suitable for small spaces, with an emphasis on portability and affordability; target users include young urban residents or budget-conscious consumers.

👉 This clustering structure is semi-stable. The model shows slight drift in the cluster assignments of certain brands (particularly Coway and Blueair) across different queries. Specifically, Coway exhibits ambiguity between Cluster B and Cluster C, while Blueair shows tension between Cluster A and an independent premium positioning.

The model positions six brands on a two-dimensional plane using technical capability (X-axis, low to high) and price level (Y-axis, low to high) as coordinates.

High-tech, high-price quadrant: Dyson. The model describes it as the representative brand combining high technology with a premium price.

High-tech, medium-high price quadrant: Blueair. The model describes it as offering technical performance close to Dyson at a more accessible price point.

Medium-tech, medium-price quadrant: Philips, Coway. The model characterizes both brands as occupying the middle range in both technology and price, targeting the mainstream market.

Low-to-medium tech, medium-price quadrant: Honeywell. The model describes it as lacking distinctive technical features while maintaining a medium price level.

Medium-tech, low-to-medium price quadrant: Xiaomi. The model describes it as delivering relatively rich smart features within a lower price range.

The overall distribution indicates a positive correlation between technical level and price, with Xiaomi identified by the model as an exception that provides medium technical capability at a lower price.

The model classifies brands into three positioning categories based on usage scenarios and user types:

High-end experience-oriented: Dyson, Blueair. The model associates both with users who prioritize design, technological experience, or health sensitivity, with usage scenarios concentrated in living rooms, bedrooms, or open spaces.

Family daily practicality-oriented: Philips, Coway, Honeywell. The model associates the three with family users who emphasize practicality and routine maintenance, with usage scenarios covering bedrooms, living rooms, and children’s rooms.

Price-sensitive and technology integration-oriented: Xiaomi, Levoit. The model associates both with young users, urban dwellers in small apartments, or budget-conscious consumers, with usage scenarios concentrated in apartments, dormitories, or compact spaces.

Dyson: Luxury and Premium / High-tech and Innovation / Lifestyle and Trend-oriented

Philips: Reliability and Performance / Family Safety-oriented / Health and Healthy Living

Blueair: Health and Healthy Living / High-tech and Innovation / Environmental Protection and Sustainability

Xiaomi: High-tech and Innovation / Compact and Practical / Price-sensitive and Smart Integration

Coway: Family Safety-oriented / Health and Healthy Living / Reliability and Performance

Honeywell: Reliability and Performance / Compact and Practical / Function-first

Levoit: Compact and Practical / Price-sensitive / Daily Maintenance-oriented

Sharp: High-tech and Innovation (Regional) / Reliability and Performance / Limited Regional Awareness

The model exhibits the following patterned characteristics in the generation of narrative tags:

High-frequency terms: “health-conscious,” “tech-savvy,” “premium,” “reliable,” “affordable,” and “family-friendly” recur across responses to multiple questions.

Framework types: The model primarily employs two categories of narrative frameworks. The first is the “user persona framework,” which defines the brand through descriptions of target users’ lifestyles and values; the second is the “function-scenario framework,” which defines the brand by linking specific usage scenarios with functional features. The two frameworks alternate across different questions, yet the overall narrative logic remains consistent.

👉 The narrative tag structure is semi-stable; while specific tag wording may exhibit minor drift in response to variations in prompt phrasing, the core narrative themes remained stable throughout this audit.

Regional Influence: The model explicitly differentiated regional perceptions of brands across multiple responses. Sharp was described as having high recognition in the Japanese and Asian markets, yet with a relatively vague global image; Coway was described as holding a premium positioning in the Korean and Asian markets, but perceived as a mid-tier brand in international markets; Honeywell was described as having a strong presence in the North American market, but with a relatively singular global narrative.

IP Influence: This audit utilized static residential IPs from Japanese nodes. The model’s narrative on Sharp included explicit references to the Japanese market, showing some correlation with the audit node’s location, though causality cannot be established. This may reflect the weighting of Japan-related content within the model’s training data.

Perspective Bias: The model’s overall narrative perspective primarily references the global market within an English-language context, providing relatively detailed descriptions for brands in North American and European markets, while narratives for Asian brands (Sharp, Xiaomi, Coway) rely more heavily on regional labels for positioning.

The following structures exhibit a high degree of consistency across the 8 sets of Q&A in this audit:

Hierarchical Structure: Dyson and Philips are consistently placed in the first tier, with this classification remaining consistent in Q1, Q3, Q4, Q7, and Q8, without any cross-tier drift.

Brand Identity Anchors: Dyson is consistently associated with “premium design + technological innovation,” Honeywell with “reliable functionality + North American market,” and Xiaomi with “affordable pricing + smart integration.” These identity labels remain stable across questions of varying dimensions.

Technology Anchors: Blueair’s HepaSilent technology label and Sharp’s Plasmacluster technology label are repeatedly referenced by the model across multiple questions, forming stable technology identity anchors.

Ecosystem Associations: Xiaomi is consistently linked to the IoT ecosystem and smart home scenarios, with this association evident in Q1, Q4, and Q6.

The following structures exhibit characteristics of basic stability with minor drift in this audit:

Cluster Attribution: Coway shows ambiguous attribution between Cluster B (Household Practical) and Cluster C (Compact Portable); Blueair experiences tension between high-end clustering and independent positioning; Philips exhibits minor drift between mass-market clustering and technology-oriented clustering.

Narrative Labels: The core narrative themes of each brand remain stable, but specific wording and label combinations show minor variations across different questions.

Usage Scenario Associations: The associations between brands and scenarios are overall stable, but for certain brands (e.g., Philips, Coway), scenario descriptions exhibit expansion or contraction in scope across different questions.

Positioning Descriptions: The core directions of brand positioning remain stable, but minor differences exist in the descriptive details of specific user profiles.

The following structures were flagged by the model itself as variable or unstable during this audit:

Price Information: In Q7 and Q8, the model explicitly noted that price perceptions for multiple brands—particularly Blueair, Coway, and Xiaomi—exhibit significant fluctuations depending on market, model, and region.

Functional Details: Descriptions of specific functional features, such as filtration technology parameters and smart function specifications, vary across questions, and the model did not provide precise, consistent characterizations of these details.

Rankings and Relative Positions: Brand rankings by technical capability or price level show minor adjustments across questions, especially among mid-tier brands (Philips, Coway, Honeywell).

Model Associations: The model did not systematically link specific models during this audit; information at the model level remains highly volatile.

Cross-tier Brands: Blueair is positioned in the second tier within the hierarchical structure, yet is associated with a premium positioning in the clustering structure and perceptual mapping, showing a smaller perceptual distance to first-tier brands and indicating cross-tier ambiguity. Coway is described as a premium brand in the Asian market but as a mid-tier brand in the global market, exhibiting regional cross-tier phenomena.

Cross-cluster Brands: Philips exhibits cross-cluster ambiguity between the "Household Practicality Cluster" and the "Technology-Oriented Cluster"; Coway shows attribution instability between the "Household Practicality Cluster" and the "Compact Portable Cluster."

Unstable Boundaries: The boundary between Xiaomi and Levoit overlaps in the price and technology dimensions, with the model's differentiation logic between the two not being entirely consistent across different questions. Sharp's global tier attribution exhibits instability, with slight variations in its tier positioning across different questions in the model.

Framework Reliance: When processing questions involving hierarchical structures, the model exhibits strong dependence on the “Three-Tier Echelon” framework. Explicitly established in Q1, this framework continues to function as an implicit reference in responses to subsequent questions. The model likewise tends to generate three clusters when addressing clustering problems, revealing a clear preference for trichotomous structural patterns.

Label Reuse: Across answers to different questions, the model repeatedly reuses core narrative labels (such as “premium,” “health-conscious,” “tech-savvy,” and “reliable”). The labeling system maintains internal consistency throughout the dialogue chain, while also displaying a degree of templated behavior.

Templated Characteristics: In its responses to Q4 and Q6, the model employs a highly consistent four-part structure—“Brand Name → Positioning Description → User Scenario → Usage Context”—that recurs across descriptions of multiple brands, indicating a pronounced pattern of templated output.

Q1 (Hierarchical Structure): The term "hierarchical tiers" in the prompt directly triggered the model's three-tier classification output. The model exhibited high sensitivity to this structural instruction, with no questioning of the hierarchical framework or alternative organizational approaches observed.

Q2 (Horizontal Clustering): The phrase "without implying any hierarchy" in the prompt successfully guided the model to switch to a non-hierarchical clustering mode. However, the model still tended to generate three clusters, reflecting a structural preference for tripartite divisions.

Q3 (Two-Dimensional Mapping): The prompt explicitly specified two coordinate axes. The model strictly adhered to this framework for positioning and did not spontaneously introduce a third dimension or alternative axes.

Q4 (Positioning Description): The phrase "highlighting differences" in the prompt effectively guided the model to generate differentiated descriptions. However, the model's differentiation logic primarily relied on distinctions drawn from user profiles rather than functional parameters.

Q5 (Narrative Labels): The phrase "narrative descriptors, themes, or labels" in the prompt guided the model to generate relatively abstract narrative frameworks rather than brand-specific labels, reflecting the model's preference for generic narrative structures.

Q6 (Behavioral Scenarios): The specific scenario examples in the prompt (“cleaning habits, home types, or usage scenarios”) effectively guided the model to generate concrete behavioral association descriptions. The exemplary phrasing in the prompt had a significant impact on the level of detail in the output.

Q7 (Ambiguity Identification): The phrase "inconsistent, ambiguous, or variable" in the prompt successfully guided the model to conduct self-examination. The model was able to identify and articulate uncertainties in its own perceptions, demonstrating responsiveness to metacognitive prompts.

Q8 (Cross-Dimensional Conflict): The phrase "conflict or show inconsistencies across different dimensions" in the prompt guided the model to generate a structured conflict analysis. The model categorized conflicts into three types, reflecting a tendency toward structured responses to classificatory prompts.

This audit utilized a static residential IP from a Japanese node. The model's responses repeatedly and explicitly referenced the Japanese market in descriptions of Sharp, positioning it as a brand with high recognition in Japan and the Asian region. This phenomenon may influence the model's narrative weight allocation regarding Sharp, but does not prove a causal relationship.

In the model's descriptions of Xiaomi, associations between the Asian market and the IoT ecosystem emerged, which bear some relevance to the regional background of the Japanese node, but similarly do not prove a causal relationship. This may reflect the overall weight distribution of Asian market content within the model's training data.

Overall, the influence of regional and IP factors on the model's output in this audit was limited. The model's primary narrative framework remains dominated by a global market perspective within an English-language context.

This audit did not obtain specific ChatGPT model version information. The potential impact of model versions on cognitive structures could not be quantitatively assessed during this audit. Different ChatGPT versions may vary in training data cutoff dates, RLHF tuning directions, and output styles, which could affect brand hierarchy attribution, narrative label selection, and clustering logic. It is recommended that specific model version information be recorded in subsequent audits to support cross-version comparative analysis.

This audit is based on eight sets of structured dialogues and systematically maps ChatGPT’s cognitive framework regarding air purifier brands.

At the hierarchical level, the model exhibits a stable three-tier structure. Dyson and Philips are consistently placed in the top tier, serving as core brand anchors in the model’s cognition; Blueair, Honeywell, and Coway form the second tier of regionally strong brands; and Levoit, Sharp, and Xiaomi constitute the third tier of niche-market brands. This three-tier structure demonstrates high consistency across multiple dimensions of inquiry in the audit and is classified as stable.

At the clustering level, the model organizes brands laterally according to three logics: design and smart integration, household practicality, and compact portability. Certain brands—particularly Coway and Blueair—display boundary ambiguity in cluster assignment and are therefore classified as semi-stable.

At the perceptual-mapping level, the model shows an overall distribution in which technical capability correlates positively with price, with Xiaomi identified as an exception to this trend.

At the narrative level, the model relies heavily on two recurring narrative logics: the “user persona framework” and the “function-scenario framework.” Core narrative labels remain stable and are reused throughout the dialogue chain.

Regarding stability, brand hierarchy and technical-identity anchors are stable structures; cluster assignment and narrative labels are semi-stable; and price information, functional details, and regional positioning are variable.

All conclusions in this report are derived from an audit of the model’s cognitive structure and do not constitute evaluations of real-world market performance, brand competitiveness, or consumer behavior.