A Brief History of Intelligence: Why the Future of AI is Human-Centric
13213_A Brief History of Intelligence.
This keynote paper by Dr. Hsiao-Wuen Hon explores the historical evolution and symbiotic relationship between Artificial Intelligence (AI) and Human Intelligence (HI). It frames AI not as a competitor to humanity, but as a sophisticated tool within the continuum of human technological advancement designed to enhance human capabilities and co-evolve with our understanding of intelligence.
TL;DR
In this insightful keynote from ICMI '19, Dr. Hsiao-Wuen Hon (Microsoft) argues that Artificial Intelligence is the latest chapter in the long history of human tool-making. Rather than a race where one must win, the relationship between AI and Human Intelligence (HI) is one of synergistic co-evolution. AI serves to augment human capabilities, while human intelligence continues to redefine itself in the presence of these new tools.
The Problem: The Myth of the "AI vs. Human" Race
The public discourse surrounding AI is often dominated by a "competition" narrative—fears of machines surpassing human cognitive abilities or replacing human relevance entirely.
Dr. Hon identifies a critical gap in this perspective: it misses the Inductive Bias of technology. Throughout history, technology hasn't just solved problems; it has changed the very definition of human potential. The "threat" of AI often stems from a lack of clarity regarding:
- Technical Boundaries: What AI can realistically achieve versus science fiction.
- Historical Context: Overlooking that AI is a tool created by HI to serve HI.
Methodology: Historical Synergy and The Evolutionary Lens
Dr. Hon proposes looking at intelligence not as a static benchmark, but as a moving target.
The Co-evolutionary Framework
The core insight is that as we develop AI, we simultaneously unlock deeper understandings of our own biological intelligence. This creates a feedback loop:
- Phase 1: Humans create tools to solve specific arithmetic or logical tasks (Logic-based AI).
- Phase 2: These tools provide data and insights that allow humans to tackle more complex abstractions.
- Phase 3: The definitions of "creativity" and "consciousness" shift as machines handle the "lower-level" cognitive duties.
Figure 1: The Keynote presentation highlighting the intersection of Multimodal Interaction and Intelligence.
Deep Insight: AI as a Multimodal Tool
As a speech technology expert and leader of Microsoft Research Asia, Dr. Hon emphasizes that for AI to truly "co-evolve," it must move beyond simple text processing into Multimodal Interaction. True intelligence integration requires:
- Speech and Natural Language Processing: Moving from dictation to true understanding.
- Contextual Awareness: Understanding the environment in which the human and AI interact.
Figure 2: Dr. Hsiao-Wuen Hon and the legacy of speech technology research.
Critical Analysis & Conclusion
The value of this paper lies in its philosophical grounding of technical development. Dr. Hon reminds us that AI is not an exogenous force acting upon humanity, but an endogenous development of our own civilization.
Key Takeaway: The "Intelligence" in AI is a mirror. If we build AI to be collaborative, we expand the boundaries of human thought. The limitation of current SOTA (2019 and beyond) isn't just compute or data, but our ability to design interfaces that allow for seamless HI-AI integration.
Constraints and Future Outlook
While optimistic, the paper subtly notes that the "appropriate use" of AI requires a technically literate society. As AI moves toward General Intelligence (AGI) concepts, the challenge shifted from "Can it work?" to "How should we work with it?" This keynote paved the way for modern discussions on Human-Centric AI and the ethical deployment of collaborative agents.