AI Advisory Training Day_April 2024 PDF

SBG Data & AI Value Tower AI Advisory Training Days 21st & 22nd of November 2024 ”Navigate the Data Jungle with Ease - SAP Data & AI” ”Navigate the Data Jungle with Ease - SAP Data & AI” ...

SBG Data & AI Value Tower AI Advisory Training Days 21st & 22nd of November 2024 ”Navigate the Data Jungle with Ease - SAP Data & AI” ”Navigate the Data Jungle with Ease - SAP Data & AI”  Let’s Forge Ahead. Agenda Day 1 1 Introduction to AI 08:45 – 09:30 with Dauren Eshenkulov 6 Use Cases from GenAI Hackathon 13:30 – 15:00 with Amit Bi Kumar 2 Introduction to GenAI 09:30 – 10:00 with Dauren Eshenkulov 7 Break 15:00 – 15:15 3 Associate with Prompts 8 AI in Cybersecurity 10:30 – 11:45 with Thuy Thu Do 15:15 – 15:45 with Thuy Thu Do 4 Winning Use Case GenAI Hackathon 11:45 – 12:15 with Marina Alej. Gonz. Petit 9 Dangers of AI 15:45 – 16:45 with Deniz Löwner 5 Lunch Break 12:15 – 13:30 10 Q&A/Closing 16:45 – 17:15 with Amit & Deniz ”Navigate the Data Jungle with Ease - SAP Data & AI” Our trainers Amit Bi Marina Dauren Vera Deniz Thu Thuy Kumar Alejandra Eshenkulov Khobotova Löwner Do Gonzalez Petit Vivek Chitransh Valerie Dylan Lott Diana Torre Krishnan Asthana Minderjahn Kadasamy ”Navigate the Data Jungle with Ease - SAP Data & AI” Q&A | Use menti 2524 9982 Data & AI – value tower in ASG Cooperation with Data & AI domain and data One factories, e.g. ATCI Data delivery Accenture India, Philippines, Riga leads Data Jürgen Data Ramona RISE data Pandrea Opportunity Schützeichel migration Data & AI migration & Pipeline services Phase II Anand Lara (to be added) Nicole Manage non Kantawala SNP Data Kraemer Milsmann migrated Service Lead AI advisory data Matthias Inga Megberg Amit Kumar Sascha Korn Westerheide Data Partner governance Data & AI People collaboration Phase II Yong Citra Siby John Andrea Holz (to be added) Bridge to Bosco Data partners (SNP, HR Business strategy & Data quality Partner architecture SAP, Syniti, etc.) Manish Gupta Rahul Singh Anand Functional Kumar & Industry Kantawala data advisory Recruiting Hans-Georg Isa & Katja Bridge to SBG Gnann Staffing & Value & Advisory Data & AI Bench Towers Phase II Management (to be added) Anja Schmitt Training Operations Rupinder Vera Ghothra Khobotova ”Navigate the Data Jungle with Ease - SAP Data & AI” AI to advance SAP Data migration AI Advisory Power Cell Purpose Main Responsibilities Initially assigned team members Identification of AI use case that can AI use case building for SAP data Amit Kumar (lead) advance the data migration ( focus on migration Korn, Sascha Generative AI) AI Upskilling Kantawala, Anand Conduct the development of use cases. If possible, as second step also Pilot development /implementations González Petit, Marina partner with existing clients for pilot Participation on Hackathon or tech Löwner, Deniz development and implementation meetups and knowledge sharing Gupta, Manish AI upskilling via trainings or external Kisic, Spasoje available training in the team Minderjahn, Valerie Understanding which data migration Klinkhammer, Daniel tools can be used for which AI Integration within and outside the value Khobotova, Vera capabilities tower Siddiqui, Huma Jump start training enhancement with Collaboration with value tower like Muratova, Kseniia AI topics data migration, data governance, SNP, data quality for use case identification Thu Thuy Do Collaboration with One Accenture Dauren Eshenkulov Collaboration with Training ”Navigate the Data Jungle with Ease - SAP Data & AI” Your time for questions Join at menti.com | INSERT 10.15 x 16.95CM OR 3.97” X 6.67” SCREENSHOT HERE use code 2524 9982 ”Navigate the Data Jungle with Ease - SAP Data & AI” Q&A | Use menti 2524 9982 Let’s Forge Ahead. 1 Introduction to AI Dauren Eshenkulov ”Navigate the Data Jungle with Ease - SAP Data & AI” ”Navigate the Data Jungle with Ease - SAP Data & AI” Welcome to the “Age of AI” 1950s – early 2000s Early 2000s – late 2010s Late 2010s – today Machines replicate Machines learn Machines create better than humans better than humans better than humans Early research on neural networks Deep learning revolution (NLP 2) Transformer architecture (2017) From early automation to Deep Blue 1 AI beats humans at Go AI achieves supremacy in language Compute capacity Data availability Compute capacity Data availability Compute capacity Data availability 500,000 > 330 million Trillion operations per Terabytes of data Trillion operations per Terabytes of data Trillion operations per Terabytes of data second (TFLOPS) produced per day second (TFLOPS) produced per day second (TFLOPS) produced per day 1. Deep Blue - was a chess-playing expert system run on a unique purpose-built IBM supercomputer. It was the first computer to win a game, and the first to win a match, against a reigning world champion under regular time controls. 2. NLP - Natural language processing, combines computational linguistics —with statistical and machine learning models to enable computers and digital devices to recognize, understand and generate text and speech. ”Navigate the Data Jungle with Ease - SAP Data & AI” Definition of AI Accenture defines AI as a set of self-learning technologies, that sense, comprehend, and act, to drive new business outcomes. These are brought together to enable machines to act with what appears to be human-like levels of intelligence. A set of self-learning technologies …that sense, comprehend and act …to drive new business outcomes Provides Insights – Has a new product Machine Learning (ML) is the core Computer Vision “sees” and interprets feature been well received? AI unlocks the technology that allows machines to mimic digital images, videos and other visual message within the data, so you can find out. the way a human learns and executes tasks. inputs. Makes Recommendations – Bought some Deep Learning is Machine Learning using Natural Language Processing interprets dumbbells recently? Your online retailer will complex neural networks, which are and generates “human-like” speech or text. use a recommendation engine to suggest you computing systems modeled loosely on the buy a protein shake. human brain, consisting of thousands or even Recommendation Engines intelligently millions of simple processing nodes that are guess what products or services a user may Makes Decisions – Have two different densely interconnected. want to buy. versions of your advertising campaigns? AI can decide which one will be more effective. Reinforcement Learning is Machine Learning Intelligent Robotics replicates human that learns through trial and error. actions and mechanical tasks. Drives New Product Design – AI can suggest alternate product designs, create prototypes and test them in virtual environments. One of the revolutionary aspects of AI is its simplicity of use, making it unique in the history of radical technologies. Humans can interact with AI through simple and natural language, whether by voice or by text, or even through images. This has one very important implication: it means the critical-mass adoption of AI is likely to take hold even faster than previous disruptive technologies. ”Navigate the Data Jungle with Ease - SAP Data & AI” Types of AI Narrow AI (Weak AI) General AI (Strong AI) Superintelligent AI ▪ Narrow AI, also known as Weak AI, refers to ▪ General AI, also known as Strong AI or ▪ Superintelligent AI refers to hypothetical AI AI systems that are designed and trained for Artificial General Intelligence (AGI), aims to systems that surpass human intelligence in specific tasks or domains. develop machines with human-level all aspects, including creativity, problem- intelligence across a wide range of tasks and solving, and social skills. ▪ These systems excel in performing a single task or a narrow range of tasks within a well- domains. ▪ Discussions surrounding Superintelligent AI defined context. ▪ The goal of General AI is to create machines often touch upon topics such as the capable of reasoning, learning, Singularity, where AI development ▪ Examples include virtual assistants, understanding natural language, and accelerates exponentially, leading to recommendation systems (Netflix, Amazon adapting to new situations in a manner unforeseeable consequences. and so on) or ChatGPT, and image similar to humans without pre-defined rules recognition algorithms. as in Narrow AI. ▪ Example include autonomous vehicle with level 5 stage, where the vehicle can act intuitively in any state or location without human intervention. ”Navigate the Data Jungle with Ease - SAP Data & AI” Approaches to AI – Symbolic AI Artificial Intelligence involves different methods and techniques for creating systems that can think and learn like humans. Symbolic AI What is it? Symbolic AI uses explicit rules and logical operations to process information and make decisions. It's like following a set of instructions to solve a problem. How does it work? The knowledge is represented using symbols, and rules dictate how these symbols are manipulated. This approach is good for tasks where the rules are well- defined and easy to express. What are the benefits? Interpretability: Provides transparency in the reasoning process, making it easier to understand how a system arrived at a conclusion. Knowledge representation: Can represent complex knowledge in a formal and structured way, allowing for easy manipulation and reasoning. Flexibility: Highly flexible and can be adapted to different domains. What are the limitations? Incomplete knowledge: Requires complete and well-defined knowledge to function correctly. In domains where knowledge is incomplete may not be effective. Scalability: Can become computationally expensive as the number of symbols and rules increases, making. Limited Adaptability: Struggle to adapt to new or unfamiliar situations that were not explicitly covered in their knowledge base or rule set. Examples: Expert Systems: These are programs designed to mimic human expertise in a specific domain, such as medical diagnosis or financial analysis. Natural Language Processing: Used to understand and generate human language, enabling applications like chatbots and language translation services. Example, if the patient reports having a fever, the system might use rule: IF patient has a fever AND patient has a cough AND patient has difficulty breathing THEN patient may have pneumonia. ”Navigate the Data Jungle with Ease - SAP Data & AI” Approaches to AI – Connectionist AI Connectionist AI What is it? Also known as neural networks, is inspired by the structure and function of the human brain. Instead of following strict rules, these systems learn from examples and adjust their behavior based on feedback. How does it work? Consists of artificial neurons organized into layers. These networks learn by adjusting the strength of connections between neurons, allowing them to recognize patterns and make predictions. What are the benefits? Can adapt and evolve with new data, enabling continuous learning and improvement. Exhibit remarkable resilience to damage or information loss, much like the human brain compensates for neuron loss. These models can generalize from specific training examples to broader applications, making them highly effective in diverse AI tasks. What are the limitations? Performance is heavily reliant on data quality and quantity. The decision-making process is often opaque and complex, making it challenging to understand or explain how specific outcomes are reached. Require substantial computational resources, which can lead to higher costs and energy consumption, posing challenges for resource-limited applications. There’s a risk of models becoming too finely tuned to training data, leading to poor performance on new, unseen data. Examples: Image Recognition: Neural networks can be trained to identify objects in images, enabling applications like facial recognition and autonomous driving. NLP: Connectionist AI powers language models capable of tasks such as sentiment analysis, language translation, and speech recognition. ”Navigate the Data Jungle with Ease - SAP Data & AI” Approaches to AI – Evolutionary AI Evolutionary AI What is it? Evolutionary AI draws inspiration from the process of natural selection and evolution. It involves algorithms that mimic biological evolution to solve problems and optimize solutions. How does it work? Evolutionary AI starts with a population of candidate solutions to a problem. Through a process of mutation, crossover, and selection, the algorithms iteratively improve these solutions over multiple generations. Evolutionary AI is often used when attempting to find an optimized solution, rather than a perfect one. What are the benefits? Robust problem-solving capacity: Exhibit resilience in handling complex optimization tasks, even in the presence of noisy or uncertain environments. Parallel processing potential: Offer promising prospects for parallel execution, enabling scalable and efficient optimization across distributed computing environments. What are the limitations? Computational overhead and resource requirements: Often demands considerable computational resources, especially for tackling high-dimensional and complex optimization problems. Sensitive to Parameter Settings: Typically have several parameters that need to be tuned, such as population size, mutation rate, and selection criteria. The performance of the algorithm can be sensitive to these parameter settings, and finding optimal values may require experimentation. Examples: In radio communications, sometimes there’s a need for designing an antenna with unusual radiation patterns for a particular mission. However, its design is not possible manually since there is an enormous number of patterns to try out. In such cases, an evolutionary algorithm comes in handy. ”Navigate the Data Jungle with Ease - SAP Data & AI” Q&A | Use menti 2524 9982 Let’s Forge Ahead. 2 Introduction to Gen AI Dauren Eshenkulov ”Navigate the Data Jungle with Ease - SAP Data & AI” ”Navigate the Data Jungle with Ease - SAP Data & AI” GenAI is much more than just a trend: it’s a reinvention “Every 14 years we get one of these Cambrian explosions. We had one around the internet in ’94. We had one around smartphones in 2008. Now we’re having another one in 2022. James Currier “ founder and venture investor who co-founded NFX ”Navigate the Data Jungle with Ease - SAP Data & AI” Why Gen AI is so relevant? $641.3 billion The market size of artificial intelligence valued in 2028 $15.7 trillion Addition to the global economy by 2030 Global enterprise adoption of Gen AI is projected to grow at a 38.1% compound annual growth rate of 38,1 % between 2022 and 2030. ”Navigate the Data Jungle with Ease - SAP Data & AI” Generative AI goes beyond traditional AI using LLM P(Cat):0.98 P(Dog): 0.98 Generative AI vs Traditional AI The AI is Generative in nature – creating numeric data, audio or visual media LLM are trained to predict the next word, therefore, to understand the context of the text, which is a realization of our language ability. We call them Foundation Model. In the text domain they are called Large Language Model Traditional AI (LLM). Source: Capgemini Research Institute analysis ”Navigate the Data Jungle with Ease - SAP Data & AI” GenAI has the capability to learn and reapply patterns of data for a wide range of applications… Inputs Outputs Text Question Answering (e.g. Wiki) Image and Videos (YouTube) Summarization Voice (Podcast) Coding Structured Data (Financial Statements) Generating content Code (GitHub, StackExchange) Automation Massive datasets Very large model trained Multimodal and emerging once, used in many ways capabilities ”Navigate the Data Jungle with Ease - SAP Data & AI” … has the capability to impact the whole organization PRODUCT DIFFERENTIATION AND INNOVATION SALES AND MARKETING CONSUMER EXPERIENCE Transforms sketches and Refine targeted marketing based Personalize purchase journeys based on descriptions into 3D product on consumer behavior and consumer profiles designs preferences Enhance customer service with chatbots Creates a foundation from which a Automate consumer segmentation that address complex inquiries creative brainstorming session can Generate tailored content using Create inclusive offerings that cater to be conducted consumer profiles and community diverse consumers Personalize offerings at scale insights OPERATIONS AND SUPPLY CHAIN IT DEVELOPMENT SUPPORT FUNCTIONS Augment warehouse operations and Testing and coding assistance for Develop training content for employees inventory management through real- increased efficiency and reduced based on role and performance time analytics manual effort Enable self-serve and automate support Avoid bottlenecks through advanced Streamline workflows by automating tasks (e.g., HR tickets, financial analysis and reporting, review of legal documents) predictive analytics and real-time repetitive tasks monitoring Generate synthetic data Source: Capgemini Research Institute analysis ”Navigate the Data Jungle with Ease - SAP Data & AI” Gen AI augments the enterprise AI continuum Diagnostic Predictive Generative What might happen How can AI help with Why did this happen? What should we do next? in the future? the execution? Advise Analyze Pattern Simulate Create Scenario Forecast Optimize Code Automate Segment Model Recommend Protect Campaign conversion and Churn risk prediction Next Best Product Hyper personalized Segment deep dive recommendations campaign messaging ”Navigate the Data Jungle with Ease - SAP Data & AI” The Main Ingredients Deep (Memory) Neural Network GPU Architectures Parallel Architecture Source: Capgemini Research Institute analysis Data ”Navigate the Data Jungle with Ease - SAP Data & AI” Generative Pre-trained Transformer Training and Inference Procedure https://jalammar.github.io/how-gpt3- https://ig.ft.com/generative-ai/ works-visualizations-animations/ 22 ”Navigate the Data Jungle with Ease - SAP Data & AI” Today’s Gen AI investments span three types of use cases that vary based on domain specificity Cross-industry Domain-specific Complex industry productivity tasks problem solving Model size Largest Large Smaller Differentiation Table-stakes Differentiated Leapfrog Buy or Build Buy Buy or fine tune Fine tune or build SDLC assistant Campaign management NOC operations optimization Examples Productivity co-pilot Customer service Service assurance Employee support Field worker companion Network and Data security ”Navigate the Data Jungle with Ease - SAP Data & AI” New ways of working: unlocked creativity Shifts in time allocation across typical design processes* 75% By 2025, marketing organizations that use AI will shift 75% of their staff’s Without gen AI operations from production to more strategic activities1 With gen AI Research Imagine Design & Test & Adjust Prototype Evaluate Gen AI can deliver Gen AI can automatically synthesized insights to generate optimized creative support rapid research Gen AI can provide a starting Gen AI can streamline highly Gen AI can automatically QA point for developing new manual design tasks to creatives against requirements concepts accelerate concepting and and best practices prototyping 1 Gartner * Illustrative ”Navigate the Data Jungle with Ease - SAP Data & AI” New ways of working: highly personalized experiences Delivering hyper-personalized experience throughout the customer journey Without gen AI Price Sensitive Self Customers are addressed as segments Purchase Segment Limited historical insights are considered Summer Price Shopper Sensitive Limited understanding of customer’s needs and interests June Relatively generic experiences Micro Segment Timeframe Journey Journey: Inspire & Money Off Self With gen AI Summer Type: Explore Options BOGO Offer Purchase Shopper Exploring Segment Customers are addressed individually Options Gold Money Off Affinity Buy One Get Actionable insights synthesized from Shopping Another $25 Off multiple sources are considered Married for Earrings Hoops and Personalized experiences are tailored Ohio Huggies! SHOP NOW Female to their specific needs and interests High Use code: SUMMER20 Price 100K Digital Sales ends 6/1* Sensitive Income Engagemt. BOGO Age: 36 Material Product Timeframe Enthusiast Personalization Personalization Segment of “1” ”Navigate the Data Jungle with Ease - SAP Data & AI” Gen AI is growing —every industry is experimenting Productivity gains* in terms of US wage bill Percentage (%) low case high case Every industry is 50% Yet