Steve Papermaster - The AI Investor

Steve Papermaster's Blog

AI Thinking: A Framework for Rethinking Artificial Intelligence in Practice

Artificial intelligence (AI) is rapidly changing how we interact with information across various fields. While its potential is vast, the diverse perspectives on what AI is and how it should be used create challenges. This article explores "AI Thinking," a novel framework proposed by Daniel Newman-Griffis in the Royal Society Open Science journal, designed to bridge these conceptual divides and guide the practical application of AI.

The Need for a New Framework: Traditional, technology-centric views of AI often fall short of capturing the complex human and contextual factors that influence its real-world application. As AI transitions from a specialized technical discipline to an everyday tool, the gap between technological advancement and best practices widens. This necessitates a framework that addresses the diverse understandings of AI across disciplines and promotes effective, ethical use.

Introducing AI Thinking: AI Thinking is a competency-based model that outlines the key decisions and considerations involved in practically applying AI. It emphasizes a holistic approach, viewing AI not just as technology but as an information methodology that shapes how we interact with information as a whole. The framework aims to connect technical processes with social practices, recognizing that AI systems are inextricably linked to their contexts.

Key Competencies of AI Thinking:

• Process (Motivating AI Use): AI adoption should be driven by specific information processing needs within a broader goal, rather than simply pursuing innovation for its own sake. This involves defining the scope of AI intervention within a process and ensuring its use is goal-driven.
• Formulation (Describing the Problem): Clearly defining the problem AI is intended to solve is crucial. This includes specifying the AI task, the desired output, and, in machine learning contexts, the training signal. Recognizing the epistemic load of problem formulation is also key.
• Tools and Technologies (Assessing AI Affordances): AI tools should be evaluated based on their technological affordances—their perceived possibilities and limitations for specific uses. This includes considering the purpose and AI paradigm, complexity and data requirements, computational needs, and strengths and limitations.
• Data (Informing AI Use): Data is not neutral. Data choices reflect perspectives and influence bias. Assessing data sources involves considering representativeness, informativeness, and reliability in relation to the AI's goals and the specific process.
• Context (Shaping AI Use): AI use is always situated within a specific context. Understanding the stakeholders, their rationales, the risks, and the measures of success is crucial for ensuring AI's effectiveness, benefits, and ethical implications.

Bridging AI Divides: The AI Thinking framework aims to reconcile different conceptualizations of AI, including:

• Linear: Focuses on computation as a process from input to output.
• Cyclical: Views AI as part of a larger process, including data collection, analysis, and action.
• Relational: Emphasizes the people, perspectives, and purposes behind AI use in sociotechnical contexts.

AI Thinking recognizes that these perspectives are complementary and equally important for a comprehensive understanding of AI in practice.

About Steve Papermaster

Steve Papermaster is a visionary AI investor, technology strategist, and entrepreneur with decades of experience in spearheading innovative solutions in AI and technology industries. He is recognized globally for his contributions to fostering technological advancement and bridging innovation with practical applications.