Beyond artificial neural nets // intelligence is simple, but..
Here are a few key points from the conversation with Noam Chomsky:
- Chomsky discussed the “Galilean Challenge” of understanding the creative capacity of human language. He believes the “inner workings” of language are largely inaccessible to consciousness and can only be studied scientifically from the outside, like any other natural phenomenon.
- On consciousness, Chomsky said we are only aware of small fragments that emerge from unconscious mental computations. To seriously study it, we need to examine the processes that allow bits to pass through to consciousness.
- On AI, Chomsky sees it as divided into science/modeling (like cognitive science) and engineering. He believes most deep learning today is brute force computation without scientific insight. True AI would need to incorporate principles like computational efficiency that characterize natural intelligence.
- Chomsky believes there is no universal scientific method, just the application of intelligence. Good education should aim to stimulate curiosity, not just transmit facts. The best teaching lays out a framework for students to creatively explore.
- On the university, he feels the physical trappings like technology don’t matter much compared to a professor able to interact and iterate with students. The key is cultivating curiosity, questioning and discovery.
- Chomsky sees current trends of ideological conformity in academia and media as displaying an irrationality and immaturity that cannot distinguish valid ideas from their speakers. He hopes we can overcome this while retaining the diversity of perspectives.
In sum, Chomsky emphasized the continuing mystery of innate human capacities like language and consciousness, the need to scientifically study the mind as a natural object, and the role of education in fostering creativity and critical thinking rather than conformity.
Any intelligence concept can be constructed from systems of two types: dictionaries (facts) and reasoning (logic) to operate data from dictionaries. Old good scientific concept, not marketing magic like emergent properties. There is an interception of logical concepts and natural intelligence properties.
“Thinking, Fast and Slow” by Daniel Kahneman is a groundbreaking exploration of the two systems that drive the way we think: System 1, which is fast, intuitive, and emotional; and System 2, which is slow, deliberate, and logical. Kahneman, a Nobel Prize-winning psychologist, draws on decades of research to uncover the cognitive biases and errors that affect decision-making.
The book delves into various aspects of human behavior, including judgment, choice, and risk. Kahneman discusses the impact of overconfidence, loss aversion, and prospect theory on our decision-making processes. He explores the concept of heuristics, mental shortcuts that often lead to systematic errors, and provides insights into how these shortcuts influence our perceptions and choices.
Kahneman also introduces the concept of “thinking about thinking” and encourages readers to become more aware of their cognitive processes. He uses real-world examples and experiments to illustrate the complexities of the human mind and offers practical implications for individuals and policymakers.
Overall, “Thinking, Fast and Slow” is a thought-provoking and accessible exploration of the psychological factors that shape human decision-making, offering readers a valuable framework for understanding their own thought processes and making more informed choices in various aspects of life.
Intelligence is simple, but we simply can’t see it.. There’s no need to attempt to understand how the brain works on a physical level and replicate it in artificial neural networks. Achieving this is currently impossible with existing technology, as our current understanding is not based on accurate models but rather on hypotheses. On the other hand, we can develop systems based on high-level research encompassing language, expert knowledge, thinking processes, and more. The fundamental concept of any intelligent system lies in knowledge representation, and natural language proves to be the most suitable form. Any other forms can be used and interpreted in natural language.
