Questions may cover: machine learning, common issues, training, evaluation, Turing test relevance, adoption, policies, the use of artificial intelligence for shopping (e.g. self-checkouts), AI hallucinations, and recent developments in large language models such as ChatGPT and Google Bard.Ā
Give a real-world example of how the concept is, has been or could be used and explain how it helps or solves a problem. Although our module touches on real-world applications, it would help if you further investigate how the concept is used and select an example which you can describe in the exam.Ā
(Dave Hullen, 2022, Source)
Machine Learning (ML) is a subset of AI that provides systems the ability to automatically learn and improve from experience without being explicitly programmed. ML focuses on developing computer programs that can access data and use it to learn for themselves.Ā Ā Ā
In essence:
AI is the broader concept of creating intelligent agents.
Machine Learning is a method to achieve AI by allowing systems to learn from data.
Example:
AI: Creating a system that can provide personalized product recommendations to customers.
Machine Learning: Developing algorithms that analyze customer purchase history, browsing behaviour, and other data to generate these recommendations.
By understanding the interplay between AI and machine learning, retailers can leverage these technologies to enhance customer experiences, optimise operations, and gain a competitive edge.
Machine Learning Ā in Retail Shopping?
Occurrence: Where does AI occur in the supply chain to improve efficiency and reduce costs? How often do AI-related issues, such as data privacy concerns, occur in the retail industry?
AI plays a pivotal role in optimising the supply chain for ecommerce businesses:
Demand forecasting: AI analyses historical sales data, market trends, and external factors to predict future demand accurately.
Inventory management: AI optimises inventory levels by balancing supply and demand, reducing stockouts and excess inventory.
Supply chain optimization: AI identifies the most efficient routes for transportation, warehouse locations, and resource allocation.
Quality control: AI-powered systems can detect defects and quality issues in products during production and shipping.
By streamlining these processes, AI significantly contributes to cost reduction and efficiency gains.
While AI offers numerous benefits, it also presents challenges:
Data privacy concerns: Retailers handle vast amounts of customer data, making data privacy a critical issue. Concerns include data breaches, misuse of personal information, and compliance with regulations like GDPR and CCPA.
Algorithmic bias: AI models can perpetuate biases present in the training data, leading to unfair outcomes for customers. For instance, recommendation systems might disproportionately suggest products to certain demographic groups.
Job displacement: Automation of tasks through AI can lead to job losses in the retail industry. Retailers must carefully consider the impact on employees and implement retraining programs if necessary.
Consumer trust: Building trust in AI systems is essential. Transparency about how AI is used, along with measures to address concerns, is crucial for maintaining customer confidence.
These issues highlight the importance of responsible AI implementation and a focus on ethical considerations.
Common Issues with AI in Retail Shopping?
AI-Powered Chatbots: These utilise machine learning and natural language processing (NLP) to understand and respond to customer queries in a more human-like manner. They learn from past interactions to improve their responses over time and can handle complex and personalised customer service tasks.
Customer Interactions: Gathered data from various sources, including in-store interactions, online purchases, customer service inquiries, and social media engagement.
Product Information: Collected data on products, including descriptions, prices, categories, and customer reviews.
Sales Data: Tracked sales transactions, including purchase history, customer demographics, and promotional effectiveness.
Cleaning and Formatting: Ensured data is consistent, accurate, and in a suitable format for analysis.
Feature Engineering: Created new features or transform existing ones to improve model performance (e.g., calculating customer lifetime value).
Machine Learning Algorithms: Chosen appropriate algorithms based on the task (e.g., recommendation systems, demand forecasting, fraud detection).
Deep Learning: Considered deep neural networks for complex tasks like image or natural language processing.
Data Splitting: Divided the dataset into training, validation, and testing sets.
Model Training: Trained the model on the training set, optimising its parameters to minimize the loss function.
Hyperparameter Tuning: Experiment with different hyperparameters (e.g., learning rate, regularisation) to improve performance.
Metrics: Used relevant metrics to assess model performance (e.g., accuracy, precision, recall, F1-score, mean squared error).
Validation Set: Evaluated the model on the validation set to identify overfitting or underfitting.
Integration: Integrated the trained model into retail systems (e.g., e-commerce platforms, point-of-sale systems).
Monitoring: Continuously monitor the model's performance in production and retrain as needed.
Continuous Learning:Ā Collecting new data, preprocessing it, retraining the model on the updated dataset, considering incremental learning techniques, and incorporating user feedback to improve the model's alignment with user expectations.
AI Training in Retail Shopping?
Image Source to Steps to implement machine learning
Model evaluation is a crucial step in the development of AI systems. It helps assess a model's performance, identify areas for improvement, and ensure it meets the desired business objectives.Ā
Remember: The choice of evaluation methods and metrics should align with the specific task and business goals. By carefully evaluating AI models, you can ensure their reliability and effectiveness.
Holdout: Splitting data into training and testing sets.
Bootstrapping: Creating multiple training/testing sets through resampling.
Cross-validation: Dividing data into folds for training and testing.
Accuracy: Overall correct predictions.
Precision: Correct positive predictions out of all positive predictions.
Recall: Correct positive predictions out of all actual positive instances.
F1-score: Harmonic mean of precision and recall.
AUC-ROC: Area under the receiver operating characteristic curve (for binary classification).
Mean Squared Error (MSE): For regression tasks.
Identifying Overfitting: The validation set helps detect if the model is learning the training data too closely, leading to poor performance on unseen data.
Fine-tuning: Based on validation results, you can adjust hyperparameters or the model's architecture to improve performance.
Accuracy: This is a general measure of how often the model correctly predicts the outcome. It's particularly useful for classification tasks like product recommendation or fraud detection.
Precision: This measures how many of the positive predictions made by the model were actually correct. It's useful when false positives are costly, such as recommending irrelevant products to customers.
Recall: This measures how many of the actual positive instances the model correctly predicted. It's useful when false negatives are costly, such as failing to recommend relevant products to customers.
F1-score: This is the harmonic mean of precision and recall, providing a balanced measure of both. It's useful when both precision and recall are important.
Mean Squared Error (MSE): This is used for regression tasks like demand forecasting. It measures the average squared difference between predicted and actual values.
Customer Satisfaction: Measured through surveys or feedback forms.
Conversion Rate: The percentage of website visitors who make a purchase.
Average Order Value (AOV): The average amount spent per order.
Return Rate: The percentage of products returned by customers.
The choice of metrics depends on the specific use case of the AI model in retail. For example, a product recommendation system might prioritise accuracy, precision, and recall, while a demand forecasting model might focus on MSE.
AI evaluation in Retail Shopping?
The Turing Test is a method of determining whether a machine can think. Proposed by Alan Turing in 1950, it involves a human evaluator engaging in natural language conversations with a human and a machine. If the evaluator cannot reliably tell the machine from the human, the machine is said to have passed the test. Ā Ā
While the Turing Test was a groundbreaking concept, its relevance in today's AI landscape is debated:
Historical Significance: It served as a foundational goal for AI research, inspiring development in natural language processing, machine learning, and other fields.
Limitations: Critics argue that the Turing Test focuses solely on conversational ability, neglecting other aspects of intelligence like problem-solving and learning.
Modern AI: Contemporary AI systems often surpass human capabilities in specific tasks, rendering the Turing Test less relevant as a comprehensive measure of intelligence.
Despite its limitations, the Turing Test remains a thought-provoking concept that has contributed significantly to the advancement of AI.
Has Retail Shop used the Turing Test method?
How does retail shopping use AI Adoption?
What policies does Retail Shop have in place with regard to AI?
Usage: How is AI employed to enhance the online shopping experience (e.g., product recommendations, personalised marketing)? How is AI utilised in brick-and-mortar stores to improve customer satisfaction (e.g., inventory management, virtual try-ons)?
Implementation: How has AI been implemented in self-checkout systems to streamline the shopping process? How can AI be implemented to optimize store layout and product placement?
AI is transforming the shopping landscape, offering a more personalised, efficient, and engaging experience. Here are some key applications:
Product Recommendations: AI analyses customer behaviour to suggest products tailored to individual preferences.
Personalised Marketing: Targeted promotions and offers based on customer data.
Virtual Stylists: AI-powered style recommendations for fashion and accessories.
Chatbots: AI-driven assistants provide 24/7 customer support and answer queries.
Visual Search: Customers can search for products using images.
Voice Shopping: AI-powered voice assistants facilitate hands-free shopping.
Inventory Management: AI predicts product demand to optimize stock levels.
Fraud Prevention: Detects fraudulent transactions to protect customers and retailers.
Supply Chain Optimisation: Improves efficiency and reduces costs.
Virtual Try-Ons: Allows customers to virtually try on products like clothes or makeup.
Augmented Reality Shopping: Creates immersive shopping experiences.
Retail Shop's use of AI?
AI is significantly enhancing the self-checkout experience. Here are some key applications:
Image Recognition: Accurately identifying products and their prices, even without barcodes.
Weight-Based Product Identification: Determining product type and price based on weight.
Fraud Detection: Identifying suspicious activity and preventing theft.
Customer Support: Providing virtual assistance for customers facing difficulties.
Queue Management: Optimizing the number of open self-checkout stations based on customer traffic.
Inventory Management: Tracking product availability and replenishment needs.
Self-Checkout in retail shop?
AI Halucinations in retail shop?
A large language model (LLM) is a type of artificial intelligence (AI) program that can recognise and generate text, among other tasks. LLMs are trained on huge sets of data ā hence the name "large." LLMs are built on machine learning: specifically, a type of neural network called a transformer model. (cloudflare)
Both Google Bard and ChatGPT are powerful large language models with similar capabilities, there are key distinctions that can influence their application in retail:Ā
Data Training: Google Bard benefits from access to Google's vast knowledge graph, potentially providing more comprehensive and up-to-date information.
Integration: Bard's integration with other Google tools (Docs, Sheets, etc.) can streamline workflows and data analysis.
Focus: ChatGPT might have a stronger emphasis on conversational abilities, while Bard could excel in providing summaries of factual topics.
While both can be used for similar tasks in retail, their strengths might be applied differently:
Product Descriptions: Bard could leverage Google Search data to create more informative and engaging product descriptions.
Customer Sentiment Analysis: ChatGPT might excel in understanding nuanced customer feedback, while Bard could analyse larger datasets for overall trends.
Predictive Analytics: Bard's integration with Google Trends could provide additional insights for demand forecasting.
The optimal choice between Google Bard and ChatGPT depends on specific retail needs and goals. Consider factors such as:
Data accessibility: If access to Google's knowledge graph is crucial, Bard might be preferred.
Integration requirements: If seamless integration with Google Workspace is essential, Bard is the logical choice.
Cost: Evaluate pricing models and feature sets to determine the most cost-effective option.
Specific use cases: Identify the core tasks where AI assistance is needed and assess which model aligns best with those requirements.
By carefully considering these factors, retailers can select the language model that best suits their business objectives.
ChatGPT, a sophisticated language model, is rapidly transforming the retail industry. Its ability to process and generate human-like text offers unprecedented opportunities for businesses to enhance customer experiences, optimize operations, and drive growth.
Developer: OpenAI
Data Source: Trained on a massive dataset of text and code, providing a comprehensive knowledge base.
Strengths: Excels in creative text formats, such as poems or scripts. Offers strong conversational abilities.
Personalised Shopping Experiences: By analyzing customer data, ChatGPT can offer tailored product recommendations, enhancing customer satisfaction.
Improved Customer Service: Providing 24/7 support, answering FAQs, and resolving issues, ChatGPT streamlines customer interactions.
Efficient Content Creation: Generating product descriptions, marketing copy, and social media content, saving time and resources.
Data-Driven Insights: Analyzing customer data to identify trends, preferences, and potential opportunities.
Optimised Inventory Management: Predicting demand and optimizing stock levels through data analysis.
Fraud Prevention: Identifying suspicious purchasing patterns to protect customers and businesses.
Enhanced Product Discovery: Enabling customers to find products through natural language search.
Virtual Assistants: Creating AI-powered assistants to guide customers through the shopping process.
Market Research: Analyzing customer feedback and market trends to inform business decisions.
Personalised Marketing: Developing targeted marketing campaigns based on customer behaviour and preferences.
By leveraging ChatGPT's capabilities, retailers can create more engaging customer experiences, optimise operations, and drive sales growth.
How is this used in your chosen retail?
Google Bard is a large language model developed by Google AI. Similar to ChatGPT, it excels at understanding and generating human-like text, making it a valuable tool for retailers.Ā Ā
Developer: Google AI
Data Source: Leverages Google Search's vast knowledge graph, providing access to real-time information and diverse data sources.
Strengths: Offers strong factual grounding and can access up-to-date information. Excels in providing summaries of factual topics.
From Persona to Platform: "The name "Bard" evoked a specific AI persona, a conversational creative partner. "Gemini" symbolises versatility and reflects Google's broader AI mission beyond a single chatbot interface."
Access to Google's knowledge graph: Bard can leverage Google's vast knowledge base to provide more accurate and informative responses.
Integration with Google products: Seamless integration with other Google tools can streamline workflows and improve efficiency.
By harnessing the power of Google Bard, retailers can create more engaging customer experiences, optimise operations, and drive growth.
How is this used in your chosen retail?
Explain in depth the impact that ONE or TWO of the following factors has Computer Security.Ā (M)
Students working at Merit level will explore ways the concept is used in the real world, and in particular how it affects end users.Ā
Students should give examples showing how it relates to some of issues to be explored
Artificial Intelligence's impact has been compared with the invention of electricity and according to the World Economic Forum, an important component of the Fourth Industrial Revolution (Hisco, David,)
positive effect
negative effect
Give examples for issues they have faced ethical and future-proofing factors
How does the use of AI in retail impact the ethical relationship between consumers and businesses, considering factors such as data privacy, data/human/algorithmic bias, and consumer trust?
Techniques that will ensure our digital outcomes remain useful in futureĀ for online shopping?
Comprehensively explain the key problems or issues related to Computer Security.Ā (E)
Choose two of the discussion prompts below to research.
Artificial Intelligenceās Biggest Challenges to Overcome
Bias
Computing Power
Integrating AI
Collecting and Utilizing Relevant Data
Man Power
Implementation Strategies
Legal Issues
Barriers and Challenges to AI Adoption in New ZealandĀ (Artificial Intelligence: shaping a future New Zealand, page 80, 81)Ā
Lack of Understanding
Lack of ScaleĀ
Risk of Being Too SlowĀ
Other Challenges in Adopting AI : lack of clear business cases , education , data issues, copyright law, privacy, security and criminal use, Defence and Military use
An AI program that recognizes speech and understands natural language is theoretically capable of understanding each conversation on e-mails and telephones.
AI systems have already started replacing the human beings in few industries. It should not replace people in the sectors where they are holding dignified positions which are pertaining to ethics such as nursing, surgeon, judge, police officer, etc.
The self-improving AI systems can become so mighty than humans that could be very difficult to stop from achieving their goals, which may lead to unintended consequences.
Source for comparison.
What are the Big Issues in your chosen Retail?
Automation is already displacing workers as we transition to a digital economy.
Artificial intelligence is also part of the digital economy, and it comes with some ethical concerns.
The Internet of Things will change entire industries, including healthcare and public works.
We should rethink jobs to include volunteer activities and other work that benefits society.
Art
Volunteer
To support economic change, we should look to portable benefits and lifetime education.
See Amazon
Source: Armstrong, 2016
Source: Sapere and Schiff Assumption, 2018 , page 94