AI’s Winograd Moment; or: How Should We Teach Machines Common Sense? Guidance from Cognitive Science
-
Darren Abramson
Abstract
In this paper, I provide a number of state-of-the-art results for Winograd- related problems by applying a public technique for zero-shot scoring with language models. I explain how I discovered this advance and my recent experience in sharing these results with the scientific community. We are on the threshold of major advances and opportunities, but also dangerous trends. A strengthened discourse on the public benefit derived from the use of computers is badly needed and starting to emerge; here, I focus on that discourse in the context of apparent sudden, giant leaps in our ability to process text with computer: natural language processing (NLP). There is no shortage of lessons for science and ethics. Fortunately, despite the vulnerable status of public University research, the prognosis for informed public debate about the sciences of machine learning and artificial intelligence remains excellent.
Abstract
In this paper, I provide a number of state-of-the-art results for Winograd- related problems by applying a public technique for zero-shot scoring with language models. I explain how I discovered this advance and my recent experience in sharing these results with the scientific community. We are on the threshold of major advances and opportunities, but also dangerous trends. A strengthened discourse on the public benefit derived from the use of computers is badly needed and starting to emerge; here, I focus on that discourse in the context of apparent sudden, giant leaps in our ability to process text with computer: natural language processing (NLP). There is no shortage of lessons for science and ethics. Fortunately, despite the vulnerable status of public University research, the prognosis for informed public debate about the sciences of machine learning and artificial intelligence remains excellent.
Kapitel in diesem Buch
- Frontmatter I
- Table of Contents V
- Acknowledgements VII
- Introduction: Affirmative and Critical Approaches to Artificial Intelligence and Human Enhancement 1
-
Part 1: Challenging “Strong AI” from the Perspective of Human Agency
- The Artificiality of the Human Mind: A Reflection on Natural and Artificial Intelligence 17
- Merits and Limits of AI: Philosophical Reflections on the Difference between Instrumental Rationality and Praxis-Related Hermeneutical Reason 33
- Experience, Identity and Moral Agency in the Age of Artificial Intelligence 51
- Outsourcing the Brain, Optimizing the Body: Retrotopian Projections of the Human Subject 79
- Life Care/Lebenssorge and the Fourth Industrial Revolution 101
-
Part 2: Examining Merits and Limits of Applied AI
- AI’s Winograd Moment; or: How Should We Teach Machines Common Sense? Guidance from Cognitive Science 127
- Passing the Turing Test? AI Generated Poetry and Posthuman Creativity 151
- Why Neuroenhancement is a Philosophical Issue 167
- The Future of Artificial Intelligence in International Healthcare: An Index 181
-
Part 3: Encounters with Artificial Beings in Film, Literature, and Theater
- Dark Ecology and Digital Images of Entropy: A Brief Survey of the History of Cinematic Morphing and the Computer Graphics of Artificial Intelligence 209
- Sentience, Artificial Intelligence, and Human Enhancement in US-American Fiction and Film: Thinking With and Without Consciousness 225
- “I, Robot”: Artificial Intelligence and Fears of the Posthuman 237
- AI on Stage: A Cross-Cultural Check-Up and the Case of Canada and John Mighton 261
- Artificial Intelligence from Science Fiction to Soul Machines: (Re‐)Configuring Empathy between Bodies, Knowledge, and Power 287
- List of contributors 309
- Index of Authors 315
- Index of Subjects 319
Kapitel in diesem Buch
- Frontmatter I
- Table of Contents V
- Acknowledgements VII
- Introduction: Affirmative and Critical Approaches to Artificial Intelligence and Human Enhancement 1
-
Part 1: Challenging “Strong AI” from the Perspective of Human Agency
- The Artificiality of the Human Mind: A Reflection on Natural and Artificial Intelligence 17
- Merits and Limits of AI: Philosophical Reflections on the Difference between Instrumental Rationality and Praxis-Related Hermeneutical Reason 33
- Experience, Identity and Moral Agency in the Age of Artificial Intelligence 51
- Outsourcing the Brain, Optimizing the Body: Retrotopian Projections of the Human Subject 79
- Life Care/Lebenssorge and the Fourth Industrial Revolution 101
-
Part 2: Examining Merits and Limits of Applied AI
- AI’s Winograd Moment; or: How Should We Teach Machines Common Sense? Guidance from Cognitive Science 127
- Passing the Turing Test? AI Generated Poetry and Posthuman Creativity 151
- Why Neuroenhancement is a Philosophical Issue 167
- The Future of Artificial Intelligence in International Healthcare: An Index 181
-
Part 3: Encounters with Artificial Beings in Film, Literature, and Theater
- Dark Ecology and Digital Images of Entropy: A Brief Survey of the History of Cinematic Morphing and the Computer Graphics of Artificial Intelligence 209
- Sentience, Artificial Intelligence, and Human Enhancement in US-American Fiction and Film: Thinking With and Without Consciousness 225
- “I, Robot”: Artificial Intelligence and Fears of the Posthuman 237
- AI on Stage: A Cross-Cultural Check-Up and the Case of Canada and John Mighton 261
- Artificial Intelligence from Science Fiction to Soul Machines: (Re‐)Configuring Empathy between Bodies, Knowledge, and Power 287
- List of contributors 309
- Index of Authors 315
- Index of Subjects 319
