Impact of Artificial Intelligence in Retail Sector
17 Nov 2023
Sheen Xavier James
Global generative AI market is forecasted to reach $53.9 billion by 2028, growing at a CAGR of 32.2% during the forecast period. Soon, robust generative AI capabilities will be a baseline requirement for any retailer that wants to stay relevant in the market. We’re only at the start of the generative AI revolution and the technology is advancing incredibly fast. From a functional perspective, the ISG report shows predictive analytics is the top use case, with 57% of all mature use cases. Code generation or DevOps (50%), data extraction and analysis (30%) and performance analysis (24%). Generative AI is potentially far more disruptive than the Industrial Revolution (a 25% improvement in productivity versus up to 80% improvement for generative AI). It represents a step change in capability because it’s both incredibly powerful and incredibly flexible.
- New ways for Customer engagement – e.g. Retail customer support. It can help in faster response to queries, explaining products, offering product recommendations etc. When used wisely, artificial intelligence yields a deeper understanding of customers across different contexts and channels and allows organizations to deepen customer relationships with truly personalized engagements. AI can read signals and sense customer’s unique intent – to purchase, upgrade, or even cancel – before they act. Real-time data, AI can serve up unique, relevant offers automatically, or guide frontline employees to make the right offer at the right time.
- Intuitive Visual Search – e.g. Customers can upload images of products they are interested in and find similar products in the retailer’s inventory.
- New ways to improve business efficiency – Generative AI excels in summarizing insights from a wide variety of unstructured and structured data sources like historical sales data, market trends, weather conditions, and social media sentiment and this makes it well suited to demand forecasting and inventory management. Retailers can now have a more accurate picture of upcoming demand, enabling them to minimize stockouts and reduce excess inventory.
- New Product design – By utilizing generative AI on existing designs and other data sources, retailers will be able to generate newer and more relevant trending designs for the brands.
- Phygital – In recent years, retailers have been focused on blending physical and digital experiences in their stores. Physical stores offer brands the opportunity to deliver impactful and delightful experiences that their ecommerce equivalents cannot reproduce or equal. Best driver of increased brand loyalty is the type and relevance of the experiences a brand can deliver. Ecommerce stores have many other benefits which physical stores may not be able to provide. Features like Click n Collect.
- Buy in store – Deliver at home, In store reviews (integrating curated online user reviews on products in a physical environment retailers), Cashier free shopping, Augmented and Virtual Reality, Location-based push notifications are some examples that generative AI promises to radically enhance.
- Merchandising – By analyzing product attributes, historical sales and customer preferences, generative AI can potentially suggest optimized store layouts and product placements. Also, it can provide personalized wayfinding experiences to store customers.
- Virtual try-ons – With generative AI able to produce extraordinarily lifelike images on demand, apparel and beauty retailers will be able to provide “magic mirror” experiences that show how different products, styles and colors would fit a customer — and even predict how that fit would change over time.
Here are some examples of early generative AI adoption in retail industry by companies to enhance customer experiences, streamline operations, and drive innovation –
● Personalized Recommendations: Amazon uses generative AI to power its recommendation engine, suggesting products to customers based on their browsing and purchase history.
● Content Generation: Alibaba’s AI copywriting tool uses generative AI to create product descriptions, reviews, and marketing content for millions of products on its platform.
● Virtual Try-On and Fitting: The eyewear retailer Warby Parker employs augmented reality and generative AI to enable customers to virtually try on glasses frames before making a purchase.
● Inventory Management: Walmart uses generative AI for demand forecasting, helping optimize inventory levels and reduce stockouts.
● Supply Chain Optimization: Zara, a fashion retailer, uses AI for supply chain optimization, including demand forecasting and inventory management, to respond quickly to fashion trends.
● Chatbots and Customer Support: H&M uses chatbots powered by generative AI to provide customer support and answer inquiries about products, sizing, and orders.
● Visual Search: ASOS , an online fashion retailer, offers a visual search feature that allows customers to upload images of clothing they like, and the AI finds similar items in its catalog.
● Fraud Detection: Shopify uses generative AI to detect fraudulent transactions on its e-commerce platform, helping protect both merchants and customers.
● Store Layout Optimization: IKEA has used generative AI to optimize the layout of its physical stores, improving the shopping experience and customer flow.
● Dynamic Pricing: Uber Eats employs generative AI to adjust food delivery prices dynamically based on factors such as distance, time of day, and demand.
● User-Generated Content Enhancement : Coca-Cola used generative AI to enhance user-generated content by generating personalized labels and messages on their bottles during a marketing campaign.
● Visual Merchandising: Farfetch uses generative AI for visual merchandising, helping curate product displays and recommendations in its luxury fashion marketplace.
While this is a promising future, there are challenges in leveraging Generative AI today, some of which are-
- Limited Understanding – there is still a lack of proper understanding and skill expertise in Generative AI, making it difficult for businesses to adopt and realize the potential.
- Investment Costs: Implementing new software, and training employees on modern technology.
- Identification of proper Use cases – Despite the top-down drive to embrace generative AI, most organizations lack the focus to identify relevant use cases which can derive outstanding business value.
- Data Quality And Bias – Generative AI requires large amounts of high-quality and diverse data to train and evaluate the models. However, collecting, classifying and processing huge amounts of data can be costly, time-consuming and challenging. If the data used to train the models is biased, the output generated will also be biased which could lead to unethical or unintended effects.
- Customer Experience: Adopting technology in areas where human interaction is needed would backfire customer experience. Proper balance between self-service options and personalized service needs to be arrived at.
- Model Interpretability And Reliability – Generative AI models are often complex and difficult to interpret and businesses fail to understand why a model makes a particular decision or prediction. This makes it critical for businesses to properly QA the output and not rely entirely on the tool. Another concern about generative AI is the potential for its misinterpretation. When generative AI can’t generate a correct answer to a question, it starts to hallucinate and invent one in a process called Artificial intelligence hallucination.
- Regulation And Ethical Concerns – It is critically important to ensure AI is used responsibly. That means setting proper controls for acquiring, refining, and deploying data. It also means addressing the impact on cybersecurity operations and managing regulatory and privacy risks. Retailers need to ensure that the AI technology they’re using is responsible by design. It is essential for organizations to understand, manage and mitigate risks resulting from AI adoption. Ethical reviews and bias screening should complement periodic risk assessments as the algorithm is evolutionary in nature, i.e., the voluminous data used to train the algorithmic models possess high velocity, heterogeneous and variability characteristics.
- Injection of Malware in software – Generative AI tools are proliferating quickly, and they are increasingly being used for things such as coding by people who don’t know how to write code. It can be a problem if the tool is compromised and embeds malware that can be detrimental to users. Hence it is imperative that a governing body needs to approve any AI tools the employee uses. Also, companies can potentially fund those tools that are relevant and safe for the employee’s job. This helps ensure better productivity, ensure quality and protect against any hostile tools.
The Artificial Intelligence Risk Management Framework (AI RMF 1.0) by National Institute of Standards and Technology (NIST) of US Department of Commerce provides a comprehensive view on how to handle the risks associated with the adoption of AI. As per this framework, AI risk management can drive responsible uses and practices by prompting organizations and their internal teams who design, develop, and deploy AI to think more critically about context and potential or unexpected negative and positive impacts. Understanding and managing the risks of AI systems will help to enhance trustworthiness, and in turn, cultivate public trust.
Characteristics of trustworthy AI systems include – valid and reliable, safe, secure and resilient, accountable and transparent, explainable and interpretable, privacy-enhanced, and fair with harmful bias managed. Highly secure but unfair systems, accurate but opaque and uninterpretable systems, and inaccurate but secure, privacy-enhanced, and transparent systems are all undesirable.
In summary, as enterprises grow more comfortable with generative AI, and use cases become more mature, we can expect more retailers to leverage generative AI to stay competitive and meet evolving customer demands.