1. Master's assignment
    1. Master on AI
      1. Three Points
    2. Student
      1. Danny Fernando Bravo López
    3. This is a report prepared for the module "Socio-economic impact of AI"
      1. "Caso de preparación: Impacto de la IA en una industria"
      2. The selected industry was Retail
    4. This report is the result of
      1. an exploration work where you can find textual references to specific topics, in such cases you will be able to find the link to go deeper (only in the original .xmind file version)
      2. an analysis to organize the information and a correlation of use cases, associated technologies and benefits
      3. personal comments according with the individual vision of Danny Bravo
    5. Date: July 2020
  2. Introduction
    1. "Every step of the retail process has the ability to be automated in a way that would increase accuracy, efficiency, and scaling of operations"
    2. An excellent reference for real current and future use cases is the Ai4 Conference
      1. Ai4 2020 is industry’s most impactful AI event. By gathering leaders of enterprise from across industry, government organizations, disruptive startups, investors, research labs, academia, associations, open source projects, media and analysts, we are creating the largest and most influential venue for AI-related idea-sharing, commerce, and technological progress.
      2. Over two days, the Retail Track at Ai4 2020 brings together business leaders and data practitioners to facilitate the adoption of artificial intelligence and machine learning technology.
      3. With a use-case oriented approach to content, our goal is to deliver actionable insights from those working on the frontlines of AI in the enterprise.
    3. Next you can find use cases organized in two categories
      1. Improve Customer experience
      2. Other use cases
  3. Improve customer experience Use cases
    1. Real-time personalization of the online shopper journey
      1. Retailers can understand shopper intent with each click & serve personalized product recommendations with the highest likelihood of engagement
      2. Role of AI
        1. Non supervised Machine learning algorithms allow to identify patterns of use
        2. Deep Learning classification algorithms allow the identification of the next action depending on the shopper's journey. This could be combined with A/B testing experimentation
        3. Recommender systems
          1. Amazon has one of the best recommendation engines on the market today, with 55% of its sales driven by its recommendations
      3. Other relevant technologies
        1. Big data is what allows the usage of big amounts of information to feed the AI algorithms
      4. Benefits of AI
        1. Machine learning helps retailers understand their customers and predict future behaviors
          1. and therefore AI helps for designing the best experience for each individual shopper
        2. Churn prevention. Finds out why customers may feel dissatisfaction.
      5. Risks associated
        1. Privacy compromised
          1. The information required to persolize the shopper's experience implies that the retailer and its associates get a deep knowledge of the individuals
    2. Personalized curation
      1. An AI styling assistant can curate looks, mood boards, outfits & collections for each shopper based on their visual style preferences
        1. One interesting solution is VueStyle
          1. VueStyle is an A.I. Styling Assistant
          2. With the Style Curation Tool, you can build theme-based curated collections for your website in quick, easy steps
          3. match your shoppers’ visual style preferences with their box preferences and picks products that you can review
      2. Role of AI
        1. Image recognition
        2. Pattern recognition
        3. Style transfer techniques using transformers
      3. Other relevant technologies
        1. BigData
      4. Benefits of AI
        1. Image Search
          1. Find products based on the inspiration moodboards of shoppers
        2. Smart Learning
          1. Get products curated based on your theme and style
    3. Analize shopper behavior in physical stores
      1. Videos of customers strolling through your shop is one of the best sources to derive insights for offline or in-store retail.
        1. You will surely spend some time defining the best technical stack, camera angles and camera position and building a map of the store. In the long run, it’s worth spending those efforts though.
        2. You would be able to streamline your business’s success by learning the micro-influences that affect how customers behave.
        3. Get the full and comprehensive image of each customer – get to know each customer’s habits, and it will be paid off by their loyalty.
      2. In a physical shopping environment, aspects of shopping behavior that may be useful to observe include:
        1. Which products draw the shoppers’ attention?
        2. Which products or which elements of the store layout cause confusion?
        3. How do shoppers navigate the store?
        4. At which times do customers prefer to shop? And is there a noticeable demographic trend? E.g. do working-age people prefer to shop in the evenings? Or in the mornings on the way to work?
        5. What emotional responses do people show throughout their shopping experience? What is the predominant emotion among shoppers at various points throughout the day and on different days of the week?
      3. Role of AI
        1. Video analytics
          1. turns regular stores into intelligent stores that have visibility into consumer behavior for optimized merchandising. Retailers can leverage video for store analytics to enable things like heat maps that show where consumers are spending the most time in the store
          2. Video analytics can also be deployed for asset protection at self-checkout kiosks and to monitor employee theft. For an average retailer, shrinkage, or the loss of inventory, accounts for approximately 1.5 to 2 percent of revenu
        2. Objects recognition
          1. Identify objetcs taken by shopper
        3. Face detection
        4. Face analysis
        5. Neural networks
          1. Identify patterns that lead to specific shipper actions
        6. Example of commercial solution
          1. Both CrowdSight Toolkit (an easy-to-use plug-and play solution) and DeepSight (SDK) come with privacy by default, allowing you to blur the recorded faces without affecting the quality of the data that you capture.
      4. Other relevant technologies
        1. High resolution cameras
      5. Benefits of AI
        1. By analyzing shopper behavior, you gain access to real-time insights and long-term trends that help you make data-driven decisions that benefit both you and your customers.
        2. arrange the products according to customer preferences
          1. by collecting data about the way that customers access the store
        3. Churn prevention. Finds out why customers may feel dissatisfaction.
        4. Reduction in theft
    4. Self attended stores
      1. The robotization of stores will result in reducing lines, lowering the number of human employees, and significant savings on operational expenses.
        1. Amazon AI has already introduced checkout-free stores
          1. Amazon Go
          2. only six to twenty human staff members are needed
          3. Just Walk Out Technology automatically detects when products are taken from or returned to the shelves and keeps track of them in a virtual cart
      2. Role of AI
        1. Computer vision
        2. Sensor fusion
        3. Deep learning algorithms
        4. Objects recognition
          1. Identify objetcs taken by shopper
        5. Face detection
        6. Behavior analysis
        7. “Detecting item interaction and movement”
          1. patented by Amazon
      3. Other relevant technologies
        1. Hi red cameras
        2. Movement detectors
        3. 1-Click shop
          1. Amazon perfected 1-click shopping at the dawn of web commerce. No company in the world has more data about buying behavior related to this type of system. Amazon also is expert in identifying products using image recognition. Combine this with the Fusion Sensors that cross confirm the new virtual “shopping cart” you create not only just by taking an item in your hand, but also by putting it back, there is actually even less of a likelihood of an erroneous transaction.
        4. Powerful App using location based services
        5. QR Code IDs
        6. Integrated Payment
        7. Multiple Sensor Technology
      4. Benefits of AI
        1. Increase sales through providing an awesome user experience
        2. Less time required to buy by a shopper
        3. Operation and Maintenance costs reduction per store
    5. In-store assistance
      1. In 2016, Lowe’s introduced LoweBot, an autonomous retail service robot designed by Fellow, in Lowe’s Stores in the San Francisco Bay area.
        1. LoweBot was able to find products in multiple languages and help customers effectively navigate the store.
      2. Pepper is the world’s first social humanoid robot able to recognize faces and basic human emotions. Pepper was optimized for human interaction and is able to engage with people through conversation and his touch screen.
        1. By Softbank Robotics
        2. Use cases
          1. Greeter
          2. Service Provider
          3. Sales Associate
          4. Loyalty management
          5. Brand Ambassador
          6. Survey conductor
          7. Retailtainment
      3. Role of AI
        1. This is a case in which there is involved almot every area of artificial intelligence
        2. Artificial vision
        3. Deep learning
        4. NLP
        5. Robotic sensors
      4. Benefits of AI
        1. Trigger curiosity, increase store traffic and attract undivided attention of shoppers.
        2. Create memorable in-store experience, transform customer journey and leverage it into a new level.
        3. Enhance product visibility, stimulate purchase and retain loyal customers.
        4. Gather comprehensive data to enrich customer base and generate shopper insights.
    6. Automatic labelling
      1. AI can create rich metadata about products
      2. Role of AI
        1. Artificial vision
        2. Big Data
      3. Benefits of AI
        1. Time reduction for labelling products
        2. Optimization of inventory state review
    7. Visual search
      1. Visual Search systems powered by Artificial Intelligence allow customers to upload images and find similar products based on colors, shapes, and patterns
      2. product categorization
        1. Lalafo sellers can just upload the image of the products they want to sell and Machine Learning retail software with computer vision would recognize it, classify it, and even suggest a price.
      3. See it - buy it
        1. users can simply snap an image of an item and will be instantly provided with all similar items that are currently available
        2. Slyce
          1. Technical details
      4. Role of AI
        1. Artificial Vision
        2. Image recognition
        3. Deel learning
        4. Big Data
      5. Benefits
        1. Increase in sales due to the ease to find the right product
    8. Chatbots to assist with customer service.
      1. AI chatbots provide an even higher level of customer service, improve searching, send notifications about new collections, and suggest similar products
        1. The base technology behind chatbots, NLP has had impressive progress last months
          1. Language model GPT-3
          2. By OpenAI
          3. Blender Bot
          4. A state-of-the-art open source chatbot by Facebook
      2. Conversica
        1. Engage
        2. Empatize
        3. Respond
        4. Conversica Sales AI Assistants Will Engage 100% of Your Leads
        5. People and Intelligent Virtual Assistants Work Together
        6. to Deliver a Personalized Touch at Scale
      3. Role of AI
        1. Natural Language Processing
        2. Big data
        3. Deep learning
      4. Benefits
        1. Improve response-time for customer attendance
        2. Offer high quality and curated data to customers
        3. Always-Available Customer Support
        4. Proactive Customer Interaction
          1. In general, companies apply a “passive customer interaction”, which means that they only respond to customers when they are contacted and not initiate the communication.
          2. In competitive businesses -especially with a remarkable percentage of millennials as customers-, none of the brands have the luxury to act passive anymore.
        5. Increased Customer Engagement
        6. Better Lead Generation, Qualification and Nurturing
        7. Cost Savings
  4. Risks associated to this kind of solutions
    1. Customer's privacy compromised
      1. Privacy is a fundamental human right
      2. As more data the companies have about their customers, bigger is the commercial opportunity they see
        1. However, that commercial opportunity may compromise the privacy of users
        2. For example, to offer the right product at the right moment, it is required to know the interests, location, acquisition capacity, loans history, and many other information about a person.
          1. Where that information reside?
          2. Who can see it?
          3. What is feasible to do with that information?
          4. What if that data is compromised/stolen?
    2. AI is going to take over jobs
      1. Automation brings big opportunities for the humanity
      2. What Jobs Will Artificial Intelligence Affect?
        1. AI could affect work in virtually every occupational group.
        2. Better-paid, white-collar occupations may be the most exposed to AI, as well as some manufacturing and agriculture positions.
        3. Business, finance and tech industries will be more exposed, as will natural resource and production industries.
        4. AI looks most destined to affect men, prime-age workers, and white and Asian-American workers.
        5. Bigger, higher-tech metro areas and communities heavily involved in manufacturing are likely to experience the most AI-related disruption
      3. The risk is not to get prepared as humanity for the change
        1. Countries can prepare their people with education
        2. However, some politicians have other priorities
      4. There are also some other key skills that humans can do really well
        1. social and emotional skill and higher cognitive skills have a lot of future promise
        2. That means an ability to work well with others, to coach, teach, and manage, but also strong problem-solving skills and critical thinking
    3. Bias
      1. The bias impact of course is still important, however, the impact in this sector affects mainly to providers, without compromising customers privacy or wellbeing
      2. There are other sectors in which the bias has a huge impact, like face recognition systems for public safety or defense.
  5. Impact of AI in the sector
    1. AI is already impacting the retail sector
      1. Smaller business are using analytics in order to make decisions about pricing, stock, campaigns, social media, and many other aspectos of the retail business
      2. Netflix and its recognized recommender systems
      3. Amazon cross selling mechanisms
    2. Short term
      1. Improve sales
        1. Better customer understanding
        2. Improving customer experience
      2. Costs reduction
        1. Logistic optimization
    3. Long term
      1. Cost reduction
        1. Automated stores
      2. Improve sales
        1. Extreme personalization
          1. Online
          2. Physical stores
        2. Massification of robotic assitants in stores
  6. Other use cases Just for reference
    1. Price adjustments
      1. Applications of AI for retail stores could help businesses set prices for their products, visualizing the likely outcomes of multiple pricing strategies. To be able to execute this, systems collect information about other products, promotional activities, sales figures, and additional data
      2. Using predictive analytics here can help to identify the best time to start decreasing or pushing prices in the other direction.
      3. AI can monitor features such as a pricing map of the market and compare demands to find out what the prices should be like
      4. Airbnb’s Aerosolve generates neighborhood polygons based on listing density, using a multi-scale k-d tree, and suggests pricing tips.
      5. Make use of coming and fading trends – optimize your price according to the current market situation, knowledge and when you can derive the maximum value from your products.
    2. Logistics and Delivery
      1. Domino’s Robotic Unit (DRU)
      2. Amazon Drones
      3. optimizing the supply chain
        1. predict how much to stock in their supermarkets
      4. Optimized Route Planning
      5. Computer Vision can help to quickly process visual information about inventory and output the relevant data about attributes and quantity into databases for further analysis.
    3. Up-selling and cross-selling
      1. predictive analytics can help here with suggestions on which goods may be combined and relate to which market segment
    4. demand forecasting
      1. We are now seeing retailers using AI and automated machine learning to operate their demand forecasting to understand the actual quantity needed today based on the demand from the customers
      2. Not only will this increase accuracy, but it increases operational efficiency, saving both time and money for the organization.
      3. Operational efficiency is absolutely key, because we're talking about an industry that is operating its business on very low operating margins
      4. Walmart's system is tracking inventor levels to alert staff when shelves need to be restocked or if fresh items have sat too long and need to be pulled.
      5. AI experts use mathematically grounded tools to predict the expected sales of an item, e.g. shoe type for a certain time period.
        1. Accurate pricing decisions are achieved by analyzing the consumer, cost and competition. With logistical and storage data, it is possible to estimate future capacity requirements, maintain availability and make accurate decisions on pricing
    5. Virtual fitting rooms
      1. Virtual fitting rooms are a great way for customers to save time and find the perfect outfit with all the elements perfectly matched — in a span of minutes! A virtual fitting kiosk from Me-Ality can scan you in 20 seconds and measure 200,000 points of your body in this period
    6. Other information for reference
      1. Use cases categories
        1. Sales and CRM Applications
        2. Customer Recommendations
          1. IBM Watson Cognitive Computing
        3. Manufacturing
        4. Payments and Payment Services
          1. Amazon Go
          2. PayPal
          3. Payment Fraud
      2. 10 uses cases Vidhya
      3. Predictive analytics
        1. Common sources for Predictive Analytics in Business