Translating AI from Concept to Reality: Five Keys to Implementing AI for Knowledge, Content, and Data
1 like753 views
The document outlines key principles and phases for successfully implementing AI in enterprises, emphasizing the importance of knowledge management, governance, and data connectivity. It presents a framework involving defining use cases, engaging domain experts, and employing iterative processes to ensure effective AI integration. The text highlights the significance of explainable AI and enriching data with contextual knowledge while continuously adapting to organizational needs.
Translating AI from Concept to Reality: Five Keys to Implementing AI for Knowledge, Content, and Data
- 1. Translating AI from Concept to Reality: Five Keys to Implementing AI for Knowledge, Content, and Data Lulit Tesfaye Partner and Division Director, Data and Information Management Enterprise Knowledge, LLC.
- 2. 10 AREAS OF EXPERTISE KM STRATEGY & DESIGN TAXONOMY & ONTOLOGY DESIGN AGILE, DESIGN THINKING & FACILITATION CONTENT & DATA STRATEGY KNOWLEDGE GRAPHS, DATA MODELING, & AI ENTERPRISE SEARCH INTEGRATED CHANGE MANAGEMENT ENTERPRISE LEARNING CONTENT AND DATA MANAGEMENT ENTERPRISE AI Clients in 25+ Countries Across Multiple Industries Lulit Tesfaye Partner and Division Director, Data and Information Management KNOWLEDGE MANAGEMENT AGILE DELIVERY PROGRAM MANAGEMENT SEMANTIC MODELING DATA & INFORMATION MANAGEMENT KNOWLEDGE GRAPHS AI Meet Enterprise Knowledge HEADQUARTERED IN ARLINGTON, VIRGINIA, USA GLOBAL OFFICE IN BRUSSELS, BELGIUM Top Implementer of Leading Knowledge and Data Management Tools 400+ Thought Leadership Pieces Published
- 3. The Opportunity: In Numbers Global business spending on AI will grow by 21.3%to reach $62.5bn by 2022. Driven by growth in five application areas: knowledge management, virtual assistants, autonomous vehicles, digital workplace, and crowdsourced data. ~ Gartner Of the businesses using AI, 91%said their ability to explain how AI arrived at a decision is critical. ~ IBM Only 8%of firms had practices that enabled them to adopt and scale AI. One of the top reasons why AI projects get stalled being the lack of relevant or usable data and having no clear strategies for sourcing the knowledge that AI requires. ~ McKinsey
- 4. Deconstructing Enterprise AI AI for the enterprise entails leveraging machine capabilities to discover and deliver organizational knowledge, data, and information in a way that closely aligns with how we look for and process information.
- 6. The Journey Toward Enterprise AI Phase I - 1960s Phase II - Early 2000s Phase III - Now Artificial Intelligence Enterprise KM AI Inception ● Logic-based problem solving From Tacit to Explicit ● KM emerges linked to economics (Druker) ● The “learning organization” and IM/KM as an “organizational resource” The IBM Watson Era ● Question answering systems ● Early image classification and processing ● Machine learning, NLP, computer vision The Era of IT and Tech Boom ● Rules-based expert systems ● The rise of the intranet ● Certification of knowledge standards and frameworks The Google and Alexa Era ● Higher computational power ● Natural language understanding and processing (BERT) ● Cloud computing & open source tools ● Increased data availability Digital Transformation Era - ● Digital transformations and KM practices ● AI and KM converge ● Collective knowledge, standards, and open data 1970s 1980s 1990s 2010s
- 7. Why Our Clients are Investing in AI FIND NEEDED INFORMATION DISCONNECTED CONTENT & DATA DYNAMISM STANDING OUT FROM COMPETITION AUGMENT KM AND DATA GOVERNANCE SECURITY AND COMPLIANCE STAFF RETENTION AND ENGAGEMENT • Provides faster and easier ways to surface content, identify customer needs, and translate them into additional products/services • Reduces the time it takes to find the right information or experts • Higher volume and dynamism of organizational content/data (both structured and unstructured) • Growing digitalization • Democratization of data access and connecting information to train AI solutions • Improved client intelligence through holistic customer conversations and efficient inquiry response • Shortened sales cycles and improved revenue through identification of large scale patterns • Relevant and timely decisions • Automates classification and labeling of content across silos • Extends scope of data lineage through machine reasoning on contextual information • Avoids additional governance costs with data standards and interoperability • Risk reduction via governance and access controls • Threat prevention by automating data capture, enrichment, investigation, archival, and retention • Provides accurate and timely responses to regulators with available and legacy data • Improved employee experience with access to collective knowledge • Predicts capacity gaps and training needs • Provides personalized training to reduce time-to-proficiency
- 8. FOUNDATIONS PROTOTYPING ENTERPRISE AI Where Our Clients are in Their AI Journey Evaluating and identifying and prioritizing AI use cases and strategy Conducting content/data gap and maturity assessments and clean-up plans Defining technical requirements, solution architectures, and plans for a proof of concept or pilot effort Providing training around AI and its foundations Operationalizing a production architecture for AI solution(s) Incrementally expanding to new use cases, data sources, departments, and solutions on a phased cadence Embedding governance and overall solution management for ongoing evolutions and releases Enterprise AI strategy and supporting organization stature Advanced architecture and integration capabilities Enterprise-wide data literacy and management training Defining organizational attributes and developing starter metadata and data models to create connections across silos for prioritized use cases Establishing their first implementable information and architecture with semantic models Launching a clickable application to demonstrate value and validate process Piloting products and tools Defining a repeatable and incremental process for future implementations Developing a plan for staffing and upskilling needs, adoption, and change management 30% 50% 11% PRE-AI 9%
- 9. The Five Keys to Implementing Enterprise AI
- 10. At the core of Enterprise AI are the foundational principles of knowledge management - people and culture, content and contextual information, governance processes - all integrated with technology. An Explainable Solution 05 Build solutions that allow users to create, modify, find, and use changing data through explainable AI Data Organization & Connectivity 03 Connect data concepts to your content (data sets, text documents, applications, etc.) People & Domain Knowledge 02 Define the knowledge important to your organization The Problem or Use Case 01 Define the problem to be solved and the overarching vision for AI Data Enrichment 04 Enrich, add, and connect data concepts
- 11. Explainable Solution 05 Build solutions that allow users to create, modify, find, and use changing data through explainable AI Data Organization & Connectivity 03 Connect data concepts to your content (data sets, text documents, applications, etc.) People & Domain Knowledge 02 Define the concepts important to your organization The Problem or Use Case 01 Define the problem to be solved and the overarching vision for AI Data Enrichment 04 Enrich, add, and connect data concepts The Problem or Use Case 01 At the core of Enterprise AI are the foundational principles of knowledge management - people and culture, content and contextual information, governance processes - all integrated with technology. A clear problem statement and definition of business value that AI will solve - one that reflects stakeholders interests. The Problem or Use Case 01
- 12. Success Factors THE PROBLEM OR USE CASE A well defined use case looks like… Construction Safety: Risk Management Recommender As a civil engineer, I will be able to see the recommended remedies for a specific construction site risk so that I can address it in my design. Goal: Optimise building design through risk mitigation/elimination based on regulatory requirements or past experience. Input Data: Standards and Procedures (SOPs), Geo data Available Source Systems: XQZ Modeling database, Access storage Output: Relevant design codes, …… For a recommendation engine: Actionable use cases are written in user story format and have the following characteristics: Identify specific end users Address a foundational business problem Garner high participant interest Communicate specific outcomes Result in a “clickable” product Are not overly complex to implement Have the required resources readily available Provide a repeatable approach for subsequent implementations What is the best approach regarding solutioning for this risk? Have we done this before? I need data to support my solution decisions. Persona Requirements
- 13. Explainable Solution 05 Build solutions that allow users to create, modify, find, and use changing data through explainable AI Data Organization & Connectivity 03 Connect data concepts to your content (data sets, text documents, applications, etc.) People & Domain Knowledge 02 Define the concepts important to your organization The Problem or Use Case 01 Define the problem to be solved and the overarching vision for AI Data Enrichment 04 Enrich, add, and connect data concepts People & Domain Knowledge 02 At the core of Enterprise AI are the foundational principles of knowledge management - people and culture, content and contextual information, governance processes - all integrated with technology. Ready access to people and domain knowledge and the supporting organizational functions. People & Domain Knowledge 02
- 14. Success Factors PEOPLE & DOMAIN KNOWLEDGE Engage Domain Experts and End Users Organizational Skill Sets and Project Teams Budget for Adoption and Governance End users and SMEs help create a rich semantic layer that captures key business facts and context for AI. Facilitate engagements through: ● Interviews and focus groups ● Design-thinking and validation sessions ● Mockups for user stories ● Gold standards ● Gathering initial feedback in modeling tools Leverage a cross-functional team, including: ● Leadership / Sponsors ● Product Owner / Manager ● Business Function Leads / Subject Matter Experts ● Technical / IT Support ● Project Delivery Team ● Governance Teams AI and knowledge governance requires collaboration across the organization. To achieve this: ● Align with the strategic goals of the organization ● Define KPIs or ROI for AI projects through an iterative process ● Engage business users from the outset and maintain their engagement and adoption ● Define a formal communications strategy and embed governance across the organization ● Start small, develop, deploy, and get feedback iteratively
- 15. Explainable Solution 05 Build solutions that allow users to create, modify, find, and use changing data through explainable AI Data Organization & Connectivity 03 Connect data concepts to your content (data sets, text documents, applications, etc.) People & Domain Knowledge 02 Define the concepts important to your organization The Problem or Use Case 01 Define the problem to be solved and the overarching vision for AI Data Enrichment 04 Enrich, add, and connect data concepts Data Organization & Connectivity 03 At the core of Enterprise AI are the foundational principles of knowledge management - people and culture, content and contextual information, governance processes - all integrated with technology. Connecting data across multiple systems enables traceable analysis, allowing you to follow information across different sources, different locations, data types, and across time. Data Organization & Connectivity 03
- 16. Success Factors DATA ORGANIZATION & CONNECTIVITY The four “C”s of data organization: Connectivity How the data entities are both related and interconnected Context The conceptual and contextual understanding of data entities Collaboration Engagement across system owners as well as between AI and humans Clean up (Complexity!) Having controls and processes in place to ensure good data quality
- 17. Metadata and Taxonomy Build a strong metadata and taxonomy foundation by defining the concepts that are important to your organization, how people talk about those concepts, and how those concepts are grouped together. You add new data concepts via taxonomy development, data entry and mapping. DATA ORGANIZATION & CONNECTIVITY How do we achieve these actions? Connecting Data Ontologies provide structure and a data schema for expanding the metadata model to create an interconnected web of information that conveys machine understandable relationships and meaning.
- 18. Explainable Solution 05 Build solutions that allow users to create, modify, find, and use changing data through explainable AI Data Organization & Connectivity 03 Connect data concepts to your content (data sets, text documents, applications, etc.) People & Domain Knowledge 02 Define the concepts important to your organization The Problem or Use Case 01 Define the problem to be solved and the overarching vision for AI Data Enrichment 04 Enrich, add, and connect data concepts Data Enrichment 04 At the core of Enterprise AI are the foundational principles of knowledge management - people and culture, content and contextual information, governance processes - all integrated with technology. Enrich your content and data with domain knowledge and context via extraction of topics and text for taxonomy enrichment, auto-tagging, and classification of key concepts in your data. Engage SMEs and knowledge engineers to set the groundwork for human-in-the-loop development and getting knowledge into a standard, machine-readable format. Data Enrichment 04
- 19. How do we achieve this? DATA ENRICHMENT Extract + Understand + Standardize + Label + Finetune + Repeat ● Training data ● Text extraction and analytics (NLP) ● Alignment of labeling with organization’s standardized taxonomy/ontology ● Auto-tagging and classification ● Accuracy, fine tuning, and iterations subject relationship object customer has product document isAbout topic Person Product Document Transaction Topic Customer
- 20. Success Factors Data Stored with Context: Store information the way people speak, with context. Data relationships are first class citizens, giving a better understanding of related data. Outcome: Understanding of the relationships between data Structured and Unstructured Information: Allow for the integration of structured and unstructured information so that users can access relevant data and content at the same time. Outcome: Connect business knowledge to data, get a 360-degree view of the organization Interoperability: Information is created based on web standards and security protocols. Making information interoperable across multiple systems. Outcome: Flexibility to move data between industry & current and future systems, minimize limitations of proprietary solutions, avoid vendor lock. Aggregation: Allow for consistent ingestion of diverse information types from sources internal or external to the organization (e.g. Linked Data, subscriptions, purchased datasets, etc.). Outcome: Handle large datasets coming from various sources, including public sources and boost knowledge discovery and efficient data-driven analytics DATA ENRICHMENT
- 21. Explainable Solution 05 Build solutions that allow users to create, modify, find, and use changing data through explainable AI Data Organization & Connectivity 03 Connect data concepts to your content (data sets, text documents, applications, etc.) People & Domain Knowledge 02 Define the concepts important to your organization The Problem or Use Case 01 Define the problem to be solved and the overarching vision for AI Data Enrichment 04 Enrich, add, and connect data concepts Explainable Solution 05 At the core of Enterprise AI are the foundational principles of knowledge management - people and culture, content and contextual information, governance processes - all integrated with technology. Combining the principles of semantic modeling with knowledge management and AI, a rich semantic layer captures key business facts and solutions and is understandable, reusable and interoperable. An Explainable Solution 05
- 22. A solution that works - Business stakeholders understand how it works & want to use it 1 Data or content that represents your business and is “NERDy” 2 Scalable practices with a portfolio of initiatives that have different milestones on the horizon 3 Closing: Success Factors EXPLAINABLE SOLUTION If we do these things right: We will have…
- 23. Lulit Tesfaye https://www.linkedin.com/in/lulit-tesfaye/ https://enterprise-knowledge.com/ Thank you!