In the 2020s, the rise of large-scale language modeling technology has led to an unprecedented acceleration in the sophistication of artificial intelligence. These machines can perform a wide range of tasks once thought only humans could solve. For example, you can write stories, create art from text descriptions, or solve complex tasks and problems that you are not trained to handle.
We asked this question to six AI experts, including James Poulos, roon, Max Anton Brewer, Robin Hanson, and Niklas Blanchard. —ed.1. In what year do you predict with 50% confidence that machines will have artificial general intelligence? That is, they will be as good as or better than most humans in all areas of learning, reasoning, or intelligence? When does it exceed?
2. What changes will this bring about in society within 5 years of the outbreak?
Anton Troynikov
AGI will be a reality by 2032. Then there will be chaos, but be optimistic. 2032. My timeline is short, but probably not as short as others. This is because there is a growing consensus that human intelligence is not particularly complex compared to other physical systems.
In robotics, there is an observation called Moravec's paradox. At the dawn of his AI research in the 1950s, cognitive tasks that were generally difficult for humans, such as playing chess or proving mathematical theorems, were considered difficult for machines as well. Sensorimotor tasks that are easy for humans, such as perceiving and navigating the world in three dimensions, were thought to be easy for machines as well. Famously, a common problem in computer vision (the field in which I have spent most of my career) was supposed to be solved in the summer of 1966.
These assumptions turned out to be fatally flawed, and the failure to create machines that could successfully interact with the physical world was one of the causes of the first AI winter, when research and funding for AI projects cooled. have become.
Hans Moravec, the man from whom the paradox is named, suggested that the reason for this is, in evolutionary terms, the relatively recent development of the human prefrontal cortex, which deals with abstract reasoning. In contrast, the structures responsible for sensorimotor functions that we share with most other higher vertebrates have existed for billions of years and are therefore very highly developed.
This also explains why we have not been able to (and to a large extent still have not) been able to infer and reproduce the sensorimotor abilities with which we evolved. Human intelligence is too immature to reason about the functioning of the sensorimotor system itself.
However, machine learning represents a way to understand the world without relying on human intelligence. Like evolution, machine learning is a purely empirical process, a general-purpose class of machines for taking in data, finding patterns, and making predictions based on these patterns. It does not make deductions and does not rely on abstractions. In fact, the field of AI interpretability exists because the way AI actually works is alien to human intelligence.
With enough data and enough computing power, AI can determine ever more complex patterns and make even more complex predictions. Since these patterns are beyond our ability to find and understand, the methods will inevitably become increasingly foreign. A concrete demonstration of this principle is the success of AI in modeling language. Linguists have been unable to provide a successful framework for automatic translation throughout the history of the field. AI used very common methods to solve problems as soon as enough data and computing became available.
Language is an expression of reason. The emulation of reason itself, by predicting what humans will reason using mechanisms foreign to reason, is not far behind. We'll get there not because AI has gotten particularly powerful, but because human intelligence is pretty weak in the grand scheme of things.
Within five years of human-level AI being created, the first upheaval will occur, followed by normalcy. Although I am generally optimistic about the future of humanity, advances in basic technology are always accompanied by upheavals. Yes, we got the printing press, but with it we also went through the Thirty Years' War.
We don't know what form this upheaval will take, but it is a fundamental upheaval as society needs to pivot around the ability to create machine intelligence on par with the average human at will. There is a high possibility that it will become something. But we get it.
Anton Troynikov has been working on AI and robotics as a researcher and engineer for the past seven years. His company Chroma improves AI by making it more interpretable.
