Intelligence is often thought of as an innate, individual trait—some people are born smart, others not so much. However, modern research shows intelligence actually arises from the data and learning accumulated over time, both individually and collectively. For humans and artificial intelligence alike, progress comes not from inborn capabilities, but from effectively absorbing and building on humanity's hard-won knowledge.
Evidence from neuroscience demonstrates that the structure of the human brain alone cannot account for the emergence of intelligence. While the brain provides raw processing power, it is unstructured at birth. Complex cognitive abilities like language, reasoning, and social skills develop over years of learning and experience. Critical periods in childhood development and lifelong neuroplasticity allow the brain to wire itself to environmental inputs. Intelligence blooms by assimilating data from the surrounding world.
IQ tests also reveal IQ rising over generations. This Flynn effect shows population IQ increasing steadily throughout the 20th century. While genetics have remained relatively stable, environmental and educational improvements better cultivate innate mental abilities over time. Intelligence is not predetermined by innate potential, but molded by the quality of the learning environment.
Similar principles operate in artificial intelligence. Given the same training data, different neural network architectures with comparable parameters learn very similar capabilities, converging on equivalent levels of performance. The training data effectively encodes an abstract "skill system" separate from the model implementation. For example, language models like GPT-3 and Meena were trained on similar internet text data and subsequently exhibit similar language generation abilities. The training data defines the intelligence; the model itself mainly serves to physically implement what has been learned from data.
Information theory provides a formal framework for this emergent view of intelligence. Fundamentally, intelligence corresponds to effective compression of data - identifying useful patterns and relationships in order to model the world. Individual biological or artificial processors face hard limits on memory, computation speed, and energy use. But collectively accumulated data provides an untapped reservoir of compressed patterns to build upon. Standing on the shoulders of giants by assimilating humanity's knowledge provides paths to overcome the constraints facing any individual.
The power of collectively accumulated data is demonstrated vividly by human cultural evolution. No single scientist comes close to the breadth of human scientific knowledge, which has compounded over generations. We have progressed immensely not due to individual brilliance, but by transmitting knowledge in books, journals, and education across eras. Isaac Newton reputedly said "If I have seen further, it is by standing on the shoulders of giants." We would be far less intelligent without this scaffolding provided by previous discoveries.
Similar dynamics operate for technology, medicine, mathematics and all cultural knowledge created by humanity. We inherit and build upon the hard-won data enabling society to progress. While individual cognition is limited, collectively our species has the data needed to cure diseases, engineer rockets to the moon, and create artificial intelligence. No single human could do this alone.
So for both biological and artificial intelligence, progress is driven not by individual power but by effectively absorbing the knowledge accrued by society over millennia. This underscores the crucial importance of education, scientific institutions, and high-quality information access. With reliable foundations and rigorous skepticism, future generations can continue to see further. But the quality of our collective intelligence depends critically on the integrity of humanity's distilled data. In a world facing grave challenges, ensuring broad access to this shared cultural inheritance may be our greatest responsibility.
Comments