AI Fundamentals: Concepts and Applications (PDF)

What is AI? Foundational Concepts U N I T 0 0 | L E SS O N 0 2 © 2025 PRODUCTSIDE, 1 LLC  DEFINITION "AI can mimic elements of...

What is AI? Foundational Concepts U N I T 0 0 | L E SS O N 0 2 © 2025 PRODUCTSIDE, 1 LLC  DEFINITION "AI can mimic elements of human behavior, not to replace us, but to help us artificial make better decisions and get more done.“ intelligence Satya Nadella, Microsoft CEO \är-ˈti-fish-əl in-ˈtel-ə-jən(t)s\ 1. : the capability of computer systems or noun algorithms to imitate intelligent human behavior “AI won't replace human 2. : a branch of computer science dealing with creativity and ingenuity. Its role the simulation of intelligent behavior in is to augment it, to help humans computers go faster and achieve more” Etymology: The term combines the Latin-derived words "artificiālis" Fei-Fei Li, Stanford AI Professor (made by skill) and "intelligentia" (understanding). It was first coined in 1956 by computer scientist John McCarthy at a Dartmouth Conference, which birthed AI. definition source: Merriam-Webster Dictionary © 2025 Productside 2  1. "... to mimic expert human decision-making to optimize our business strategy." 2. "... to simulate customer interactions to improve “I have this our customer service experience." 3. “… to generate accurate messaging content to problem. accelerate marketing campaigns.” 4. "... to replicate the diagnostic skills of top medical professionals to assist in healthcare." So, how might we 5. "... to emulate the cognitive processes of a financial analyst to enhance our investment strategies." use AI to …” 6. "... to imitate human-like teaching and coaching for personalized education. © 2025 Productside 3  AI: Historical Perspective AI has a long history of solving problems dating back to the 1950’s. Since then, many branches and approaches have developed and flourished. ChatGPT and other solutions like it were built on decades of previous work. © 2025 PRODUCTSIDE, 4 LLC  Computer Vision Healthcare – Assists in Computer Vision is AI analyzing medical images like technology that mimics human MRIs. vision by analyzing visual data from images and videos. Autonomous Vehicles – This technology assists in the Detects objects, pedestrians, automation of visual tasks, and traffic signals. image-based decision-making, and problem-solving across Manufacturing – Ensures various industries. quality control by detecting product defects. © 2025 Productside 5  Expert Systems Expert Systems are AI Customer Service – Help schedule orders, applications designed to route customer requests, and solve emulate the decision- problems. making abilities of human experts by utilizing Telecommunications – Make decisions specialized knowledge and about network technologies and reasoning. maintenance of existing networks While highly effective in specific, rule-based areas, Mechanical Engineering – Troubleshoot they necessitate extensive complex electromechanical machinery manual coding. designs. © 2025 Productside 6  Machine Learning Healthcare – Aids drug discovery Machine learning allows computer and personalized treatments based systems to enhance their on patient DNA. performance by learning from data instead of relying on explicit Retail – Product and pricing programming. recommendations (e.g., Amazon, Netflix). ML developed from Expert Systems by enabling models to learn from featurized data instead Recruiting – Helps job boards match of being programmed with specific candidates with job opportunities. rules. © 2025 Productside 7  Neural Networks Healthcare – Enhances medical image Neural Networks are a type of analysis, diagnosis, drug discovery, and machine learning algorithm personalized treatments. inspired by the human brain, consisting of interconnected Financial Services – Powers fraud nodes that learn to recognize detection, risk assessment, trading, and patterns through training. customer behavior prediction. They were created to overcome the limitations of traditional Robotics – Enables advanced machine learning by perception, motion planning, and automatically learning adaptive control for intelligent complex patterns from raw, autonomous systems. high-dimensional, and unstructured data. © 2025 Productside 8  But what about Large Language Models and Generative AI? © 2025 Productside 9  Understanding LLMs & GAI A creative branch of AI: Makes new things: text, images, music, etc. Made up of multiple AI: Deep learning, neural networks, & NLP © 2025 Productside 10  Large Language Models Copywriting – Can be used to improve original style and voice (AI21 LLMs are AI systems designed Wordspice) to understand, generate, and assess human language Code Generation – things like Amazon by identifying patterns within CodeWhisperer not only create code, large text datasets. but create SQL queries They play a crucial role in creating natural language Writing – Used to generate text for products and services documentation, or write a short story focused on text generation and like Alexa Create analysis. © 2025 Productside 11  Generative AI In conversational AI, ChatGPT generates answers, assists with tasks, Generative AI involves deep- and creates text content. learning models capable of producing high-quality text, images, and other content. In data generation, Mostly AI creates fully anonymous synthetic data from This technology fosters innovation your original data. by automating creativity, customizing user experiences, and facilitating new methods of In media creation, SORA produces content creation. audio, video, and multimedia content. © 2025 Productside 12  Is the noise about Fine- Tuning, RAG, and Agentic AI Hype? © 2025 Productside 13  Fine-Tuning Retrieval Augmented Generation (RAG) Definition: Customizes an AI model Definition: Extends an AI model by with specific data to improve domain- retrieving real-time external data specific tasks. without modifying it. Use Cases: Use Cases: Media/marketing: Generate Legal/medical: Ensure accurate, specialized, creative content with real-time responses using up-to- consistent style and tone. date, factual information. Customer support: Tailor Rapidly changing environments: responses to align with specific Access the latest data for news, brand messaging or product details. finance, or stock analysis. © 2025 PRODUCTSIDE, © 2025 Productside 14 LLC  Understanding Agentic AI Agentic AI is a sophisticated form of artificial intelligence that can autonomously perform tasks, make decisions, and interact with users using its acquired knowledge. By combining technologies such as LLMs, NLP, and Rules- Base Systems, Agentic AI acts as an independent decision- making "agent" across various contexts. © 2025 Productside 15  Characteristics of Agentic AI We need to Build a We need to Figure out cross the bridge cross the river how river! ASAP!!!!!! An AI system that adapts high dynamically, exhibiting agency in tasks and workflows while operating within accountability Accountabili guardrails. ty Hope an agent is working on the river problem lo w low Autonomy high © 2025 Productside 16  How does Agentic AI Solve problems? Healthcare – Diagnostic tools assist doctors in making faster, more accurate decisions. Agentic AI autonomously makes decisions and Finance – Investment strategies interacts, streamlining optimize portfolios based on real-time processes across industries financial data. by learning and adapting. Autonomous Vehicles – Self-driving cars make real-time driving decisions, improving road safety and efficiency. © 2025 Productside 17  ACTIVITY Brainswarm AI Topics Everybody, all at once, go crazy and drag and drop topics from the right—or write your own— into the following categories: Throwdown AI topics that interest me: topics you find exciting or intriguing. GOAL AI topics I need for work: topics crucial As a group, let's kick off our learning journey by to solving problems in your role. seeing which key AI topics we understand versus those that confuse us versus those we need and AI topics I find confusing: topics you those we don't. don’t fully understand or need more clarity on. AI topics I want to avoid: topics that feel TOOLS irrelevant or unhelpful to your work. Mural TIME 20 minutes © 2025 Productside 18  Disruptive Opportunities Leverage AI to innovate, optimize strategy, and stay competitive. © 2025 PRODUCTSIDE, © 2025 PRODUCTSIDE 19 LLC  What does ‘Disruptive Innovation’ mean to you? © 2025 Productside 20  DEFINITION “Disruptive innovation doesn't always come with shiny gadgets. It's about rethinking how to solve problems and disruptive innovation deliver value in ways that fundamentally shift the competitive \dis-ˈrəp-tiv ˌi-nə-ˈvā-shən\ landscape.” ~ noun Clayton M. Christensen (1952-2020), Management Professor & 1. : An innovation that upends markets by offering Author simpler, cheaper, or more accessible solutions, shaking up the status quo 2. : When newcomers outsmart giants with fresh "The best innovations are often the thinking that delivers unmet value, reshaping how we define progress. simplest ones. Sometimes, all it takes is to reimagine a familiar problem and Etymology: The term was popularized by Harvard Business School come up with a new way to solve it. scholar Clayton Christensen in his 1997 book "The Innovator's That's the essence of disruptive Dilemma. innovation."~ Liu Qing (a.k.a. Jean Liu), definition sources: Wikipedia & "The Innovator's Dilemma" president of Didi Chuxing © 2025 Productside 21  Timeline of Disruptive Innovation Creating Value A TIMELINE Generative The Web Mobile AI LLMs, Deep Learning, HTML/http Internet Everywhere NLP Brands interact directly Intimate experiences via This is why you are here with customers, eCommerce, etc. App Ecosystems today!​ 1990 2025 Algorithms, personalizatio Bitcoin, crypto, NFTs, etc​ n Security, authentication, not Targets advertising to so sure about NFTs though users, influences customers​ … Social Blockchain Networking © 2025 Productside 22  How Solving Problems Drove AI Innovation Introducing Duolingo Max, a Apple Plans AI- The hype we Generative AI and the Walmart uses AI to learning Powered Health read: Fraud Examiner predict demand experience Coach powered by GPT-4 Lack of practical Accessibility to Hard to predict what High rates of fraud with application in lessons health coach that customers want/little Core problem: traffic that is too high for and limited feedback isn't time consuming time for consumer manual monitoring on incorrect answers and is affordable research Increase learning Examines workout Customers always speed and and diet info to Reduces credit card fraud by find what they are comprehension with recommend holistic AI solved by: efficiently examining billions looking for, and fresh answer explanations health improvement of transactions per day goods are delivered and roleplaying plans and guide you faster. conversations to do it. © 2025 Productside 23  How do we problem solve at the Speed of Change? l tia nen xpo e Martec’s law [mär-ˈteks-ˌlȯ] noun : a principle that states that technology changes exponentially, but organizations change much more slowly (logarithmically). CHANGE This is Reset where Product M Trajectory anagement ge is needed​ an l ch gica lo hno tec logarithmic nge niz a tio nal cha orga © 2025 Productside TIME 24  What is AI? Disruptive Opportunities Agenda AI Topic Throwdown Case Study Stay Calm Feel their Stake Out Stand on Unpack Build on Socialize and Focus Pains, an AI the Assumptio Tiny Bets an on the Learn their Playing Shoulders ns & for Big Outcomes Problem Gains Field of Giants Unknowns Wins Blueprint Space © 2025 Productside 25  What are some problems AI is SOLVING CREATING for you today? © 2025 Productside 26  The Case Study An Overview of the Case Study that Drives our Lessons © 2025 Productside 27  A HiPPO walks into backlog planning and bellows: "Drop what you're doing and add Generative AI to our product pronto!" How do you respond? ** This image was generated with AI (put your response in the chat channel) © 2025 Productside 28  Grab a towel and stay focused © 2025 Productside 29  TL;DR … we need to understand Before we talk about what problem we want to how AI can add value solve, and to your product… who we want to solve it for. © 2025 Productside 30