where W_Q, W_K are learned query and key matrices, and d_k is the dimension of the key vectors. Visualizing these attention maps reveals which sustainability factors most influence recommendations.

Counterfactual Explanations

Counterfactual explanations answer: "How would this recommendation change if certain sustainability attributes were different?" Given a recommendation y = f(x), we seek the minimal perturbation δ such that:

For a cotton t-shirt recommendation with high water usage, the system might show that switching to organic hemp would reduce water consumption by 60% while maintaining similar style attributes.

Implementation Considerations

Practical deployment requires balancing explanation fidelity with computational overhead. Layer-wise relevance propagation (LRP) offers efficient explanations for DNNs by redistributing the output prediction backward through the network:

where R represents relevance scores, z_ij are the activations, and ϵ stabilizes numerical computation. This allows real-time explanation generation without significant latency penalties.

Traditional recommendation systems often prioritize accuracy metrics such as precision, recall, or mean squared error (MSE). However, in sustainable fashion recommendation, these metrics alone fail to capture the broader impact of the system. A holistic evaluation must incorporate environmental, ethical, and user-centric dimensions.

Environmental Impact Metrics

The carbon footprint of recommended items can be quantified using life cycle assessment (LCA) data. Let c_i represent the CO₂ equivalent emissions per unit of item i. The average carbon footprint of recommendations over N users is:

where R_u denotes the set of items recommended to user u. Similarly, water usage W and land use L can be modeled with analogous formulations.

Ethical Sourcing Compliance

Let s_i ∈ [0,1] represent a normalized ethical sourcing score for item i, aggregating factors like fair labor practices and material traceability. The ethical compliance E of a recommendation set is:

where R is the union of all recommendations. This requires integrating certified supply chain data into the feature space.

Diversity and Novelty

To avoid homogenization and promote circular fashion, we measure the Shannon entropy H across material categories m ∈ M:

where p(m) is the probability mass of material m in recommendations. Novelty can be assessed via the average item age Ā in inventory days, discouraging fast-fashion overstock.

Behavioral Shift Measurement

The system's long-term efficacy depends on inducing sustainable purchasing habits. This is captured through the repeat sustainability ratio ρ:

where S is the set of sustainable items and 𝕀 is an indicator function tracking consecutive sustainable purchases over T transactions.

Multi-Objective Optimization

These metrics form a vector f = [Ā, H, E, 1-ρ] that can be optimized using Pareto frontiers. The recommendation problem becomes:

subject to accuracy constraints, where α are tunable weights reflecting stakeholder priorities. This framework enables explicit tradeoff analysis between sustainability and traditional performance metrics.

The environmental impact of fashion recommendations can be quantified through a multi-dimensional assessment framework that integrates lifecycle analysis (LCA), carbon footprint estimation, and water usage metrics. For an AI-driven recommendation system, this requires embedding sustainability scores into the optimization objective while maintaining relevance to user preferences.

Lifecycle Assessment Integration

Lifecycle assessment evaluates the environmental burden of a garment across raw material extraction, manufacturing, transportation, usage, and disposal phases. The total impact I of a recommended item can be modeled as a weighted sum:

where E_k represents the environmental impact metric (e.g., kg CO₂ equivalent) for lifecycle phase k, and w_k are phase-specific weights derived from industry benchmarks. For textiles, key phases include:

• Material production (e.g., cotton farming, synthetic fiber extrusion)
• Manufacturing (dyeing, cutting, sewing energy use)
• Transportation (distance and mode-specific emissions)
• Use-phase (washing, drying energy)
• End-of-life (landfill, recycling, or incineration impacts)

Carbon Footprint Estimation

The carbon footprint C of a recommendation can be computed using activity-based accounting:

where a_i is the quantity of activity i (e.g., kWh of electricity), f_i is the emission factor for that activity, and C_transport and C_eol capture transportation and end-of-life emissions respectively. Emission factors should be sourced from databases like Ecoinvent or industry-specific LCA studies.

Water Usage Calculation

Water consumption W is particularly critical for natural fiber production. The water footprint can be decomposed into:

where blue water is surface/groundwater consumed, green water is rainwater, and grey water represents pollution-adjusted usage. For cotton, typical values range from 2,700 to 10,000 liters per kg of fiber depending on irrigation practices.

Multi-Objective Optimization

The recommendation system must balance environmental impact with traditional metrics like purchase probability. This can be formulated as a constrained optimization problem:

where R(x) is the predicted relevance score, I(x) is the environmental impact, and λ ∈ [0,1] controls the trade-off between sustainability and relevance. The feasible set X_feasible may include constraints on price, style compatibility, or other business rules.

Impact Normalization and Comparison

To enable cross-category comparisons (e.g., cotton t-shirt vs. synthetic jacket), impacts should be normalized per wear or per garment lifetime. A useful metric is impact per wear (IPW):

where n is the estimated number of wears before disposal. This accounts for durability differences between materials - a longer-lasting item may have higher absolute impact but lower IPW.

Real-World Data Integration

Practical implementation requires integrating with supply chain databases that provide:

• Material composition percentages
• Country-of-origin for manufacturing steps
• Transportation routes and modes
• Brand-specific sustainability initiatives

APIs from platforms like Higg Index or OpenApparelRegistry can provide this data at scale. For missing data points, machine learning models can estimate impacts based on similar products with known footprints.

Experimental Design for Sustainable Fashion Recommendation Systems

To assess the behavioral impact of AI-driven sustainable fashion recommendations, controlled experiments must employ a mixed-methods approach combining quantitative metrics with qualitative insights. A longitudinal A/B testing framework is optimal, where:

• Group A receives conventional fashion recommendations
• Group B receives sustainability-optimized recommendations
• Group C (control) receives no recommendations

The key dependent variables include:

where weights α, β, γ are derived through conjoint analysis of user preferences. The experiment should run for a minimum of 6 months to account for seasonal fashion cycles.

Behavioral Metrics and Psychographic Segmentation

Advanced clustering techniques reveal distinct user segments based on their response to sustainability cues:

where x_i and y_i represent normalized behavioral vectors across:

• Click-through rates on sustainability labels
• Dwell time on environmental impact information
• Purchase conversion rates for eco-friendly items
• Post-purchase satisfaction surveys

Neurocognitive Response Measurement

Cutting-edge studies employ EEG and eye-tracking to measure subconscious responses to sustainability information presentation. Key findings indicate:

Counterfactual Impact Analysis

The true causal effect of recommendations requires potential outcomes framework:

where Y(1) is the sustainability outcome with recommendations and Y(0) without. Doubly robust estimation combines propensity score weighting with outcome regression to minimize confounding:

Field Experiments in E-Commerce Platforms

Large-scale deployments on platforms like ASOS and Zalando demonstrate:

• 23.4% increase in sustainable purchases when recommendations include real-time environmental impact comparisons
• 17.8% higher retention rates among users who engage with sustainability filters
• 9.2% reduction in return rates for items recommended with detailed sustainability information

The most effective interventions combine:

1. Personalized sustainability scores aligned with user values
2. Dynamic explanations of recommendation rationale
3. Social norm comparisons within peer groups

Real-Time Recommendation Systems

Integrating AI-driven sustainable fashion recommendations into e-commerce platforms requires a real-time, low-latency inference pipeline. The recommendation engine must process user interactions (clicks, views, cart additions) and environmental data (carbon footprint, material sustainability scores) to generate personalized suggestions. A common approach involves a hybrid architecture combining collaborative filtering and content-based filtering:

Here, μ represents the global average rating, b_u and b_i are user and item biases, p_u and q_i capture latent factors from collaborative filtering, while c_u and f_i encode content-based features like material composition and supply chain transparency.

API-First Integration

E-commerce platforms typically expose RESTful APIs or GraphQL endpoints for third-party integrations. The recommendation service must:

• Subscribe to real-time event streams (user browsing, purchase history)
• Query product catalogs with sustainability metadata
• Return ranked recommendations within < 100ms latency

A well-designed integration leverages asynchronous processing for sustainability scoring while maintaining synchronous responses for critical user-facing recommendations.

Data Synchronization Challenges

Product catalogs in fast-fashion e-commerce update frequently (often hourly). The AI system must handle:

Where ΔP represents the model parameter updates needed to adapt to new inventory, L is the loss function incorporating sustainability metrics, and B is the batch of newly added products. Continuous learning requires careful versioning of product embeddings to prevent recommendation drift.

Personalization at Scale

Sustainable fashion recommendations must balance personal preferences with ecological impact. This multi-objective optimization can be formulated as:

Where U(x) represents user utility, S(x) is the sustainability score, and α controls the trade-off. The constraint matrix A encodes business rules like inventory availability and price ranges.

Edge Caching for Performance

To meet strict latency requirements, recommendation systems employ edge caching strategies. The cache invalidation policy must account for:

• User preference dynamics (decay factor γ ≈ 0.85)
• Sustainability score updates (typically daily batch updates)
• Inventory changes (real-time invalidations)

This requires implementing a hybrid cache with TTL-based expiration for static features and event-driven invalidation for dynamic attributes.

AB Testing Framework

Measuring the impact of sustainability recommendations requires careful experimental design. The hypothesis test compares conversion rates between control (standard recommendations) and treatment (sustainability-aware) groups:

Where p̂₁ and p̂₀ are conversion rates in treatment and control groups respectively. The test must account for network effects and seasonal trends in fashion purchases.

Mobile applications leveraging AI for sustainable fashion recommendations integrate multi-modal data processing, real-time decision-making, and user behavior modeling to optimize eco-conscious purchasing. These systems typically employ hybrid architectures combining collaborative filtering, content-based filtering, and reinforcement learning to balance personalization with sustainability metrics.

Architectural Components

The core pipeline consists of three interconnected modules:

• User Preference Engine: Processes historical interactions via transformer-based sequential recommendation models. The attention mechanism weights items based on temporal relevance:

where Q, K, and V represent queries, keys, and values matrices derived from user interaction sequences, and d_k is the dimension of key vectors.

• Sustainability Scoring Module: Computes environmental impact scores using lifecycle assessment (LCA) data through fuzzy logic systems. Each garment's score S combines material, production, and transportation factors:

where w_i denotes normalized weights for n sustainability criteria, and μ_i(x_i) represents membership functions for fuzzy variables like water usage or carbon footprint.

• Multi-Objective Optimization: Balances user preferences against sustainability scores using Pareto frontier analysis. The recommendation utility function U for item j is:

where P_j is the predicted preference score, S_j the sustainability score, and α a tunable parameter controlling the trade-off.

Real-Time Implementation Challenges

Mobile deployment introduces latency constraints requiring:

• Quantized neural networks with 8-bit integer precision (INT8) for on-device inference
• Edge caching of sustainability databases using Bloom filters for efficient membership queries
• Differential privacy mechanisms when processing sensitive style preferences

Modern frameworks like TensorFlow Lite implement per-channel quantization during model conversion:

converter = tf.lite.TFLiteConverter.from_saved_model(saved_model_dir)
converter.optimizations = [tf.lite.Optimize.DEFAULT]
converter.target_spec.supported_types = [tf.int8]
quantized_model = converter.convert()

Evaluation Metrics

Performance is measured through:

• Sustainability-Adjusted Hit Rate (SAHR): Proportion of recommended items with both high click-through rates and LCA scores above threshold τ:

where R_u^k denotes top-k recommendations for user u.

• Carbon Efficiency Ratio (CER): Measures reduction in estimated supply chain emissions per successful recommendation compared to baseline browsing behavior.

Scaling AI-driven sustainable fashion recommendation systems presents several technical and operational hurdles. One primary challenge is the data sparsity problem, where user-item interaction matrices are highly sparse due to the niche nature of sustainable fashion. Traditional collaborative filtering methods struggle with this, as the lack of sufficient user preferences leads to poor generalization. Hybrid approaches combining content-based and collaborative filtering can mitigate this, but they require extensive feature engineering and domain-specific knowledge.

Computational and Infrastructure Constraints

Training deep learning models for fashion recommendation demands significant computational resources, particularly when processing high-resolution images or multi-modal data (e.g., text descriptions, material sustainability scores). The energy consumption of large-scale training runs contradicts the sustainability goals of the system itself. For instance, training a transformer-based model like BERT for fashion text analysis can emit up to 1,400 lbs of CO₂, undermining the environmental benefits of the recommendations.

Here, E_training is the total carbon emissions, P_GPU is the power consumption per GPU hour, t_i is the training time per epoch, and C_carbon is the carbon intensity of the energy source.

Dynamic Inventory and Cold-Start Problems

Sustainable fashion brands often have rapidly changing inventories due to limited-edition or seasonal items. This introduces a cold-start problem for new items with no historical interaction data. Techniques like meta-learning or few-shot learning can help, but they require careful tuning to avoid overfitting. Additionally, the lack of standardized sustainability metrics across brands complicates feature alignment. For example, one brand's "organic cotton" rating may not be comparable to another's due to differing certification standards.

User Adoption and Behavioral Resistance

Even with accurate recommendations, user adoption is hindered by cognitive biases such as the "fast fashion mindset," where consumers prioritize low cost and trendiness over sustainability. A/B testing studies show that users often ignore sustainability labels unless they are prominently displayed alongside price and style information. Reinforcement learning can optimize the presentation of recommendations, but this requires real-time feedback loops that are difficult to implement at scale.

Regulatory and Ethical Considerations

Deploying AI recommendations in the EU under the Digital Services Act requires transparency in algorithmic decision-making. This conflicts with the proprietary nature of many recommendation models, particularly deep learning systems where explainability is limited. Techniques like SHAP (Shapley Additive Explanations) or LIME (Local Interpretable Model-agnostic Explanations) can provide post-hoc explanations, but they add computational overhead and may not fully satisfy regulatory requirements.

Greenwashing—misleading claims about environmental sustainability—poses a significant challenge for AI-driven fashion recommendation systems. Advanced techniques are required to ensure recommendations are genuinely sustainable rather than superficially labeled as such. This involves multi-modal data verification, supply chain transparency analysis, and dynamic trust scoring.

Multi-Modal Data Verification

Traditional recommendation systems rely on product metadata, which can be easily manipulated. A robust approach integrates:

• Material composition analysis via spectral imaging or chemical fingerprinting
• Supply chain blockchain records verifying ethical sourcing
• Lifecycle assessment (LCA) databases with verified environmental impact metrics

The verification pipeline can be formalized as a constrained optimization problem where the recommendation score S is weighted by authenticity factors:

Where R is the traditional recommendation score, V_i are verification metrics, and w_i their respective weights learned through:

Dynamic Trust Scoring

Brand sustainability claims require continuous validation through:

• Time-decayed credibility models where older certifications lose influence
• Cross-referenced NGO databases tracking labor violations
• Satellite imagery analysis of manufacturing facilities

The trust score T(t) at time t follows an exponential decay with reinforcement:

Where γ_k represents verification events at times t_k, and λ controls decay rate.

Case Study: Detecting Synthetic Cotton Claims

A 2023 implementation by the Sustainable Apparel Coalition used hyperspectral imaging to distinguish organic from conventional cotton with 94% accuracy. The classifier architecture:

Trained on 50,000 fiber samples, the model reduced false sustainable claims by 37% in pilot deployments.

Optimizing sustainable fashion recommendations requires solving a multi-objective problem where cost constraints compete with environmental impact metrics. The core challenge lies in formulating an objective function that weights these competing factors appropriately for different user segments.

Mathematical Formulation of the Trade-off

The recommendation system must minimize a composite objective function combining price and sustainability scores:

Where:

• x represents a fashion item in the recommendation pool
• P(x) is the price of item x
• S(x) is the sustainability score (higher is better)
• α and β are weighting parameters (α + β = 1)
• Min/max values are normalized across the inventory

Dynamic Weight Adjustment Based on User Behavior

The weighting parameters should adapt to individual users through reinforcement learning. We model this as a contextual bandit problem where:

Where:

• u_t represents user context at time t
• φ is a feature mapping function
• w are learnable parameters
• σ is the sigmoid function bounding α between 0 and 1

Inventory-Aware Recommendation Constraints

Real-world systems must account for limited sustainable inventory. We formulate this as a knapsack problem with probabilistic constraints:

Where:

• r_i is the combined score for item i
• c_i is the sustainability cost (e.g., carbon footprint)
• B is the target budget
• ε controls the strictness of the constraint

Implementation Considerations

Practical systems require efficient approximate solutions. We employ:

• Dual decomposition to separate the multi-objective problem
• Thompson sampling for the bandit problem
• Sample average approximation for the chance constraints

The complete system architecture involves:

1. Real-time feature extraction from user sessions
2. Periodic retraining of the weighting model
3. Batch optimization of the recommendation pool
4. Online adaptation via bandit feedback

Generative Adversarial Networks for Material Synthesis

Recent advances in generative adversarial networks (GANs) have enabled the synthesis of novel sustainable materials through computational design. A conditional GAN architecture, such as StyleGAN3, can model the relationship between material properties (e.g., tensile strength, biodegradability) and molecular structure. The generator G learns to produce synthetic material representations that minimize an environmental impact metric E, while the discriminator D enforces physical plausibility:

where λ controls the sustainability trade-off. Recent work by Ecovative Design has demonstrated GAN-generated mycelium-based textiles with 78% lower carbon footprint than conventional cotton.

Quantum Machine Learning for Supply Chain Optimization

Quantum annealing architectures (e.g., D-Wave 2000Q) are being applied to solve NP-hard supply chain optimization problems in fashion. The Hamiltonian for sustainable logistics routing can be formulated as:

where d_ij represents distances between nodes, c_ij^CO2 captures transport emissions, and α, β are Lagrange multipliers. Current implementations show 40-60% reductions in carbon emissions compared to classical solvers for problems with >500 nodes.

Neural Radiance Fields for Virtual Try-On

Neural radiance fields (NeRF) enable photorealistic virtual garment visualization without physical samples. The continuous scene representation F_Θ maps 5D coordinates (3D position + 2D viewing direction) to color c and density σ:

Differentiable rendering computes the pixel color Ĉ(r) for ray r through volumetric integration:

where T(t) = exp(-∫_tn^tσ(r(s))ds) models light attenuation. Current systems achieve 94% accuracy in predicting fabric drape behavior, reducing return rates by 35%.

Blockchain-Enabled Provenance Tracking

Hyperledger Fabric implementations combine convolutional neural networks (CNNs) with Merkle-patricia tries for immutable material provenance. Each transaction T_i contains a CNN-embedded material fingerprint h_θ(x_i) and sustainability metadata:

The consensus protocol verifies each block's mean environmental impact score S_B against industry thresholds:

with weights w₁, w₂ derived from life cycle assessment databases.

Differential Privacy in Recommendation Systems

ε-differentially private algorithms protect user data in sustainable fashion recommendations. The privacy loss is bounded by adding Laplace noise Lap(Δf/ε) to gradient updates in federated learning:

where sensitivity Δf is computed over adjacent datasets. Current implementations maintain 85% recommendation accuracy while guaranteeing (ε=0.5, δ=10^-5)-differential privacy.

• PDF An Intelligent Mobile Application to Recommend Clothing Using Machine ... — sustainable themes, effectively raising awareness among high-end consumers but neglecting the broader market and practical sustainability engagement. Slow Low Fashion aims to address these shortcomings by leveraging AI for personalized fashion recommendations, thus reducing overconsumption and waste. It democratizes sustainable fashion
• Sustainable digital fashion in a metaverse ecosystem — Academics and practitioners have long regarded the sustainable development of the fashion industry as a key issue. Growing attention on the environmental costs of fast production ( Niinimaki et al., 2020 ) has spurred research on more sustainable models like slow fashion ( Henninger et al., 2016 ; Sener et al., 2019 ) and the circular economy ...
• Implementation of Digitalized Technologies for Fashion Industry 4.0 ... — 5. AI in Fashion. In this section, we discuss the significance of AI in the fashion industry for multiple applications such as prediction of health-related issues of elders, patients, and children, fashion trend forecasting, and dress recommendation based on environmental parameters. 5.1. Trend Forecasting and Dress Recommendation
• PDF The Impact of Artificial Intelligence on Personalized Fashion ... — the overall fashion consumption experience, while also considering the historical context of the control group's interactions with fashion e-commerce websites without AI. 4.3 Variables 4.3.1 Independent Variable The primary focus of this study is the integration of artificial intelligence (AI) in personalized fashion recommendations
• A Systematic Review of Reimagining Fashion and Textiles ... - MDPI — Artificial intelligence (AI) is revolutionizing the fashion, textile, and clothing industries by enabling automated assessment of garment quality, condition, and recyclability, addressing key challenges in sustainability. This systematic review explores the applications of AI in evaluating clothing quality and condition within the framework of a circular economy, with a focus on supporting ...
• The Future of Artificial Intelligence in Fashion: Innovations ... — The aim of this research article is to investigate the prospective role of artificial intelligence (AI) in the fashion business. This article seeks to offer a complete overview of how AI is altering the fashion industry by analysing the present state of AI applications, identifying new trends and developments, and evaluating the difficulties ...
• AI in Sustainable Fashion Market Size & Industry Growth 2030 — [150+ Pages Report] The global Artificial Intelligence in Sustainable Fashion Market size was valued at USD 1.35 billion in 2022 and is projected to expand at a CAGR of 35.5%, reaching a value of USD 18.5 billion by 2030.
• (PDF) Smart Fashion: A Review of AI Applications in the Fashion ... — The fashion industry is on the verge of an unprecedented change. The implementation of machine learning, computer vision, and artificial intelligence (AI) in fashion applications is opening lots ...
• PDF ARTIFICIAL INTELLIGENCE: AI IN FASHION AND BEAUTY E-COMMERCE - Theseus — Chapter 5 Research results: quantitative results based on a survey about AI in general and satisfaction with AI features. Chapter 6 brings out the future of AI in fashion and beauty e-commerce. The sub-questions and the thesis's central query are addressed in Chapter 7. The chapter
• (PDF) Advanced Fashion Recommendation System for ... - ResearchGate — The fashion industry is rapidly expanding and playing a critical role in driving global economies. Due to this ever-growing industry, application of computer science is rising rapidly to solve ...

• Artificial intelligence and sustainability in the fashion industry: a ... — The fashion industry often falls short of sustainability goals, but contemporary technological advancements offer a wide range of tools to address this issue. Artificial Intelligence (AI) has emerged as a particularly promising ally in promoting sustainability in fashion. This literature review explores how AI can contribute to the fashion industry's sustainability, highlighting its ...
• AI And The Future Of Sustainable Fashion - Sigma Earth — The Solution: AI in Sustainable Fashion . The fashion business is at a critical juncture of change. Change is looming, and the rise of artificial intelligence (AI) is a possible catalyst for sustainable Fashion.AI continues to have immense impacts on the fashion cycle concerning design, production, and even sales, with the possibility of environmental sustainability.
• The Role of Artificial Intelligence in Enhancing Sustainability in the ... — Sustainability remains a challenge in the fashion sector, yet recent technological progress provides an array of solutions, with Artificial Intelligence (AI) taking a lead role. This review paper aims to explore how AI is being applied to foster sustainability in the fashion industry, while identifying its advantages and limitations. Through a qualitative analysis of 16 peer-reviewed articles ...
• AI in Fashion: How AI Can Revolutionize Sustainability of Fashion? — Conclusion. AI represents a game-changing technology that is driving the fashion industry towards a more sustainable future. By leveraging data analytics, machine learning, and automation, AI enables brands to design products, source materials, report sustainability metrics, and optimize operations in ways that were previously unimaginable.
• A Systematic Review of Reimagining Fashion and Textiles Sustainability ... — Artificial intelligence (AI) is revolutionizing the fashion, textile, and clothing industries by enabling automated assessment of garment quality, condition, and recyclability, addressing key challenges in sustainability. This systematic review explores the applications of AI in evaluating clothing quality and condition within the framework of a circular economy, with a focus on supporting ...
• Fast Fashion's Fate: Artificial Intelligence, Sustainability, and the ... — Artificial Intelligence is defined as "a system's ability to interpret external data correctly, to learn from such data, and to use those learnings to achieve specific goals and tasks through flexible adaptation" (Kaplan & Haenlein, 2019, p. 17).While AI is not a new technology, it remains a subject of great interest in the fashion and financial sectors.
• AI's Role in Crafting Sustainable Fashion: A Green ... - Refabric — From the selection of sustainable materials to the optimization of production processes, AI algorithms are helping fashion designers make environmentally conscious choices. By analysing data on the ecological impact of materials and suggesting alternatives, AI ensures that each step in the fashion design aligns with the principles of ...
• (PDF) Fashion Sustainability in the AI Era: Opportunities and ... — The study delves deep into the fashion industry's evolving landscape, centering on sustainable practices and the novel role AI is playing in this shift, particularly from a marketing standpoint.
• (PDF) Artificial intelligence and sustainability in the fashion ... — Our findings demonstrate that AI has diverse applications in different aspects of the fashion industry, enhancing sustainability efforts in supply chain management, creative design, sales and ...
• (PDF) Artificial Intelligence in Sustainable Fashion Marketing ... — The rise of artificial intelligence (AI) has vastly transformed our economy and is now beginning to herald a sea change in various industries, including fashion. This paper explores how AI is ...

• Perception of Innovative Usage of AI in Optimizing Customer ... - MDPI — The research was designed to contribute to scientific efforts in exploring the attitude of fashion stakeholders towards AI and its use in attaining sustainability in fashion industry. Although the role of AI in Fashion has been studied before, the aim of this research is to challenge and analyze the attitudes towards sustainable fashion of both stakeholders and consumers. The research ...
• AI Enhancing Sustainability In Fashion Brands - heuritech.com — Key takeaways Artificial intelligence technology offers a sustainable solution to fashion, reducing overall inventory levels by 20-50% and improving labor conditions in the fashion industry. Fast fashion poses a serious threat to the planet, with global textile production more than doubling in the last 15 years and total greenhouse gas emissions from this at 1.2 billion tonnes annually. Brands ...
• AI in Sustainable Fashion Market Size & Industry Growth 2030 — [150+ Pages Report] The global Artificial Intelligence in Sustainable Fashion Market size was valued at USD 1.35 billion in 2022 and is projected to expand at a CAGR of 35.5%, reaching a value of USD 18.5 billion by 2030.
• PDF Is AI the future of Sustainable or Green Fashion: A Review Report - JETIR — For fashion brands, AI and ML can be a game-changer. It will allow for faster customer service, better user experiences in the store environment and more sustainable business practices.
• A Systematic Review of Reimagining Fashion and Textiles ... - MDPI — This study used quantitative methods to compare the types of datasets used and drew on qualitative methods to thematically analyze the studies and identify the potential applications of AI for supporting the sorting and classification of textiles and garments to promote a more circular fashion industry.
• Chatbots and AI in Fashion Industry | SpringerLink — The chapter on "Chatbots in the Fashion Industry" provides a comprehensive educational resource on the evolving role of chatbots in the fashion sector and the impact of AI. It uses practical case examples to stimulate critical thinking on technology issues, user experience dynamics, and ethical aspects.
• PDF Ai Based Personalized Outfit Recommender — The AI-based Personalized Outfit Recommender, the focal point of this innovative project, represents a groundbreaking leap into the future of fashion curation. This system, driven by state-of-the-art Convolutional Neural Network (CNN) algorithms, aims to redefine the way individuals engage with their wardrobes and make informed choices for ...
• Digital humans in fashion: Will consumers interact? — The adoption of digital humans by fashion business can bring about several advantages. AI-based digital humans can know your taste better and can make clothing exactly to your measurements (McDowell, 2020a). This is likely to result in fewer returns and thus help create a more sustainable future.
• Perception of Innovative Usage of AI in Optimizing Customer Purchasing ... — The research considers the views of consumers, industry professionals and company shareholders on the role AI plays in pursuing ideas of Sustainable Fashion.
• (PDF) Smart Fashion: A Review of AI Applications in the Fashion ... — The implementation of machine learning, computer vision, and artificial intelligence (AI) in fashion applications is opening lots of new opportunities for this industry.