دانلود کتاب برنامه‌های پیشرو هوش مصنوعی سازمانی – بهینه‌سازی تیم‌های هوش مصنوعی برای خلق ارزش

This is a handbook for all those who wish to lead a team of specialists using artificial intelligence (AI) to create real and lasting business value in an enterprise. It is not a book on AI neither as a scientific branch of mathematics nor as a technology topic in software form. It is more a book on leadership in the concrete context of AI. AI projects fail to deliver business value mainly for three reasons: 1. Users do not want to use the software application that brings the AI model to them for a variety of reasons. 2. Users want to use it but cannot efficiently or effectively do so because the interface, either to the user or to the business process, is clunky in some way. 3. The AI application solves a problem no one has or rather solves only part of the real problem or a different version of it. In essence, this book is a manual on how to avoid these three leading causes of death of AI projects. While avoiding them will not guarantee success, it will improve your odds by a factor of ten in individual projects. In addition to the view on individual projects, this book discusses the reality of an enterprise AI program and team having a portfolio of projects with overlapping technological requirements and conflicting priorities. As your overall enterprise goal is not necessarily success in every project but rather the overall success of the AI program consisting of many projects, success is now virtually assured. The root principle is a relentless focus on business value. AI should not be pursued because it is flashy, modern, or the latest hype but used surgically where it makes sense as a tool to solve a real problem. Value can be achieved only if users use the product you build. The product needs to be fully integrated into the wider process for this to work and you need to build trust among the user group. Trust on the one hand is user confidence in the output of the AI application and on the other hand it is trust that the goal of the AI program is not laying off the very users you are dealing with. For this, the AI team must develop strong change management, empathy, and training skills. AI scientists want to improve their models and hit accuracy records. While that is good, done is better than perfect. The AI team must embody standards for when an outcome is good enough and then move on because the goal is not scientific excellence but business value. The key insights of the book are these: 1. Start with a clear, transparent, and concrete vision of what the AI program will deliver for the whole company— an enterprise AI strategy. 2. Structure a team and an operating model that can deliver on that strategy. 3. Identify use cases by looking for challenges currently faced by the business that can be solved meaningfully by AI and assemble these into a prioritized portfolio. 4. Create a comprehensive infrastructure for the deployment of the required AI models for the project portfolio using as many synergies as possible and taking into account all dependencies such as data access and quality as well as cybersecurity. The goal is not to deploy one use case after another but to deploy the entire portfolio! 5. For each project, carefully analyze business needs, write them down, and get business stakeholders to formally sign off on them. Use agile project management to conduct the projects with frequent stakeholder involvement. 6. Focus relentlessly on user experience and change management as these two elements together are necessary and sufficient conditions for successful adoption and therefore value generation. 7. Consistently emphasize the goal of value generation by carefully measuring cost and benefit throughout the project and watching project risks such as data quality. 8. Institutionalize central AI governance processes to ensure ethical and responsible use of AI throughout with appropriate risk management and guardrails. 9. Centralize the management of AI vendors such as research institutions, product vendors, and services companies to limit duplication of efforts and ensure high-quality standards. 10. Finally, create an enterprise-wide AI community through events and education fostering mutual learning and innovation.