Geoffrey Hinton (Google Scientific Advisor) – Heidelberg Laureate Forum (Jan 2020)


Chapters

00:00:17 Early Origins of a Scientist's Curiosity: A Penny's Uphill
00:05:53 Hinton's Reflections on Early School Experiences
00:12:29 Childhood Experiences and Questioning Authority in Education
00:17:53 A Mathematician's Journey from Disdain to Understanding
00:24:10 Interest in the Workings of the Brain
00:27:26 Philosophy vs. Science: A Methodological Perspective
00:33:50 Existential Psychoanalysis and the Intellectual Development of Geoffrey Hinton
00:36:29 Child Development: Beyond the Scope of Early Psychological Theories
00:41:15 AI Development: From Symbolic to Connectionist Approaches
00:48:26 Geoffrey Hinton's Early Explorations in Neural Networks
00:50:51 The Journey of a Maverick Scholar: From Academia to Cognitive Science
00:54:20 A Lone Voice in Machine Learning: Geoffrey Hinton's Unique Perspective

Abstract

Geoffrey Hinton: A Journey of Intellectual Curiosity and Defiance

Abstract:

This article delves into the remarkable journey of Geoffrey Hinton, a trailblazer in artificial intelligence, from his early inquisitiveness piqued by a penny moving uphill on a bus seat to his revolutionary work in AI. It charts his struggles and inspirations during his formative years, his rebellion against conventional norms, and his pivotal contributions to the field of AI. The narrative follows an inverted pyramid structure, commencing with the most significant aspects of his life and gradually delving into detailed accounts of his experiences and ideologies.

Geoffrey Hinton’s intellectual odyssey, marked by an early fascination with a penny’s unusual movement, encompasses a series of transformative experiences and insights. From grappling with religious and political norms in his educational settings to challenging the status quo in the field of artificial intelligence, Hinton’s path was anything but conventional. His unwavering pursuit of comprehension, extending beyond science to the broader expanse of human cognition, culminated in monumental contributions to AI, notably his advocacy for neural networks and backpropagation, in defiance of prevailing academic norms.

Early Curiosity and Education:

Geoffrey Hinton’s journey into the world of inquiry and rebellion commenced at age five, with a simple observation during a bus ride home. The peculiar movement of a penny on a velvet bus seat, seemingly defying gravity, ignited a spark of curiosity that would define his approach to life and learning. His educational years were fraught with challenges: isolation due to his family’s political beliefs, a struggle with religious doctrines in a non-religious upbringing, and the influence of distinct personalities, like his math teacher S.T.P. Wells and an art teacher with sadistic tendencies. These experiences, though diverse, molded Hinton’s rebellious nature and relentless questioning spirit.

Early Mathematical Struggles:

Despite his later mathematical prowess, Hinton struggled with mathematical concepts, particularly functions, during his high school years. He had a concrete mindset and preferred mechanical analogies to grasp abstract ideas. It wasn’t until he ventured into programming as a graduate student that functions finally made sense to him.

The Importance of Deep Understanding:

Hinton emphasizes the paramount importance of deep understanding, striving to comprehend concepts thoroughly enough to construct them from scratch. He applies this approach to various disciplines, encompassing psychology and emotions.

Cambridge University: An Unexpected Turn:

Hinton’s academic trajectory took an unanticipated turn when he abruptly departed from Cambridge after a month due to the overwhelming pressure. He spent time in London, engaging in odd jobs and delving into literature, ultimately developing an interest in architecture.

Architecture and the Reality of Practice:

Hinton reapplied to Cambridge to pursue architecture but discovered the mundane reality of the profession during a summer internship. Disillusioned by the practical constraints, he switched back to natural sciences with the support of his tutor, highlighting the flexibility of the Oxbridge approach.

The Quest for Deeper Understanding:

Despite excelling in science, Hinton found himself at odds with the conventional educational system, frequently questioning its methodologies and teachings. His dissatisfaction with concepts he did not fully grasp compelled him to delve deeper into abstract concepts, seeking to comprehend the underlying principles. This trait was evident in his studies, where he transitioned from natural sciences to psychology at Cambridge, only to find it lacking in theories and proper experimentation. His quest for a more substantial understanding of intelligence and cognition ultimately steered him towards artificial intelligence.

Cambridge’s Psychology Course:

Hinton’s education in psychology at Cambridge focused on scientific aspects, including rat studies and signal detection theory. He felt dissatisfied, finding the course lacking in insights into human psychology, particularly psychoanalysis.

Existential Psychoanalysis Tutorial:

Hinton approached his tutor, seeking permission to attend weekly tutorials in existential psychoanalysis in London. The college surprisingly agreed to fund these tutorials, demonstrating their commitment to a liberal education. While Hinton appreciated the psychoanalyst’s beautiful girlfriend, he struggled to grasp the concepts of Husserl and Heidegger.

Hinton’s Intellectual Development:

Hinton’s psychology course had limited impact on his intellectual growth. Existential psychoanalysis sparked his interest, but he acknowledges the difficulty in understanding certain concepts.

Psychology’s Lack of Theoretical Foundation:

Hinton criticizes psychology’s lack of a proper theory that explains behavior. He finds it lacking compared to physics, which has theories that explain phenomena.

Behaviorist Psychology’s Limited Scope:

Hinton describes an experiment he conducted with children to study their attention to shape and color. The experiment aimed to determine if children’s attention shifted from color and texture to shape as they grew older.

Unexpected Behavior in the Experiment:

During the experiment, a bright five-year-old child pointed out a red circle among yellow triangles and a yellow square. The child reasoned that Hinton had made a mistake in painting the circle red, indicating an understanding of intentions and errors.

Complexity Beyond Existing Theories:

This behavior demonstrated a level of reasoning that was beyond the scope of the theories in psychology at the time. Hinton compares the theories to a stepladder, which is inadequate for reaching the complexity of human behavior.

Disenchantment with Psychology:

Hinton expresses disappointment with psychology’s naive use of the experimental method to test inadequate theories. He finds the field’s approach ineffective in explaining complex behaviors like the one he observed in the experiment.

Transition to Artificial Intelligence:

Hinton’s transition to AI was not straightforward. Initially intrigued by child language development, he critiqued prevailing theories and conducted his own experiments. His move to the AI center in Edinburgh marked the beginning of his significant contributions to the field. Despite encountering skepticism and opposition, particularly regarding neural networks and backpropagation, Hinton persisted. His unwavering dedication to his research, often against the counsel of his advisors and established academic norms, demonstrated his unwillingness to conform.

Early Influences and Chomsky’s Theory of Language:

Hinton’s initial academic pursuits were interrupted by a year of carpentry work. He later became involved in a project studying child language development, influenced by Chomsky’s theory of innate syntactic structures in language.

Exploring Child Language Development:

The project involved observing young children in Bristol and recording their language using radio microphones and tape recorders. Researchers aimed to analyze children’s language samples to understand the acquisition of complex grammatical structures.

Tags as a Window into Grammatical Structures:

Tags, such as “isn’t he” or “aren’t they,” were studied to assess children’s ability to express intricate grammar. One memorable example involved a parent telling a child, “Santa don’t give you no toys if you don’t talk proper, isn’t he?”

Transition to Artificial Intelligence:

Inspired by the potential of artificial intelligence, Hinton pursued a PhD in the field in 1972. He joined the newly established AI center in Edinburgh, which focused on symbolic AI.

Lighthill’s Report and the Symbolic AI Debate:

Sir James Lighthill’s report criticized the symbolic AI approach, arguing that it was inefficient and lacked a solid foundation in how the brain processes information. Lighthill’s analysis sparked controversy within the AI community, with some defending symbolic AI’s capabilities.

The Impact on Hinton’s Career:

As a graduate student during this period, Hinton was influenced by peer pressure to believe in symbolic AI. Following Lighthill’s report, Hinton’s advisor left Edinburgh for Sussex, prompting Hinton to transfer as well.

Navigating the AI Landscape:

Hinton’s experiences shaped his views on AI and led him to question the limitations of symbolic AI. He continued his research in AI, eventually contributing to significant advancements in the field, including deep learning and neural networks.

Advisor and Dissertation Topic:

Geoffrey Hinton’s official advisor was Jim Howe, but it was understood that Hinton would pursue his own interests. Hinton wanted to work on neural networks and how they learned but struggled to find significant improvements over existing knowledge.

Hinton’s advisor had previously worked on neural networks but switched to symbolic AI, impressed by Terry Winograd’s thesis using symbolic methods for natural language commands. Despite his advisor’s attempts to persuade him otherwise, Hinton remained stubborn and determined to pursue his interests in neural networks.

Hinton’s advisor recognized his stubbornness and tenacity, traits common among graduate students. An instance is recalled where Hinton’s advisor presented an idea to him, but Hinton politely declined, stating that he had his own ideas to work on.

Hinton’s advisor exhibited tolerance and understanding, respecting Hinton’s independence and allowing him to pursue his own research interests.

Neural Networks and AI:

Hinton’s work in neural networks, a concept that existed since Turing’s time but was not prominent, became a cornerstone of his career. He resisted the shift towards symbolic AI and persevered with his research, despite numerous challenges, including the limited job opportunities in AI in Britain following the Lighthill Report. His relocation to America, where he collaborated with like-minded individuals such as David Rommelhardt, further solidified his position in the field. Hinton’s commitment to applying backpropagation to practical problems, such as speech and object recognition, distinguished his work and contributed significantly to the advancement of AI.



Geoffrey Hinton’s journey exemplifies the transformative power of intellectual curiosity and the courage to challenge conventional wisdom. His early experiences, marked by a blend of isolation, rebellion, and inspiration, shaped his unique approach to learning and understanding. His contributions to artificial intelligence, particularly in neural networks and backpropagation, have left an indelible mark on the field. Hinton’s story is not just one of scientific achievement but also of a relentless pursuit of knowledge, driven by a deep desire to comprehend the world in its most intricate forms.


Notes by: ChannelCapacity999