<My Opinion>
I majored in computer science. So, I was confident that I know quite a lot about the history of artificial intelligence's transformation. However, this article that i've read provided me totally a new thought that I've never imagined about the artificial intelligence.
I was already aware that the concept of artificial intelligence was motivated from how neurons work to transfer a stimulation. When a stimulation that exceed threshold is sensed by any part of a human's body, a sparse set of neurons among gazillion neurons fires action potential. Human can feel the stimulation as the action potential reaches all the way through an axon terminal. AI researchers developed the concept of AI based on how a stimulation is transfered.
However, Jeff Hawkins -a neuroscientist and tech entrepreneur- proposes slightly different term of how the study of AI should aspire. Hawkins proposes that the study of AI should be relavant with neuroscience. Specifically, he says, AI models should have intelligence like human's brain does. Below is how human's brain does crucially to have intelligence.
1. Learn by moving(so-called "embodiment") - sense a particular circumstance by moving eyes, hand and etc.
2. Sensory input gets taken up by tens of thousands of cortical columns.
3. Continous learning - not to forget previous stuff.
4. Structure knowledge using reference frames - relating sensory input to understand the particular circumstance.
The above steps ought to be adopted to develop AI to have intelligence like human's brain, he says. He also emphasized that the purpose of AI is not just solving a particular situatain -such as Alphago wins human in Go-, but preserving knowlege of human, so AI can preserve ourselves for a time and a place we don't know yet.
This article is quite innovative in that it suggests a new direction for AI research to pursue.
<ChatGPT 첨삭>
As a computer science major, I was confident that I had a solid understanding of the history and evolution of artificial intelligence. However, after reading this article, I was introduced to a perspective I had never considered before.
I was already aware that the concept of artificial intelligence was inspired by how neurons transfer stimuli. When a stimulus exceeding a certain threshold is detected anywhere in the human body, a small subset of neurons among billions fires an action potential. This action potential travels through the axon terminal, allowing humans to perceive the stimulus. AI researchers initially developed AI by modeling this process of neural signal transmission.
However, Jeff Hawkins—a neuroscientist and tech entrepreneur—proposes a slightly different approach to AI development. He argues that AI research should be more closely aligned with neuroscience. Specifically, he believes that AI models should mimic the way human brains process intelligence. He outlines the following key functions that are crucial for human intelligence:
- Learning through movement ("embodiment") – Sensing and interacting with the environment by moving the eyes, hands, etc.
- Processing sensory input – Information is processed through tens of thousands of cortical columns.
- Continuous learning – Retaining past experiences rather than forgetting them.
- Structuring knowledge using reference frames – Relating sensory input to contextualize and understand experiences.
According to Hawkins, these principles should be incorporated into AI development to create models that exhibit true intelligence, similar to the human brain. He also emphasizes that AI’s purpose is not merely to solve specific problems—such as AlphaGo defeating humans in Go—but rather to preserve human knowledge. In doing so, AI could serve as a means of safeguarding human existence for an unknown time and place.
This article is particularly thought-provoking because it suggests a new direction for AI research—one that aims to make AI more human-like in its cognitive abilities.
<ChatGPT 평가>
끝!
