A study published last July revealed that forgetting, a concept often considered negative in our lives, can actually greatly benefit AI learning capabilities. This not only challenges our traditional understanding of learning and memory but also offers a unique perspective on how AI might influence our physical and cognitive environments. Artificial neural networks, powering most advanced AI systems today, mirror the complex network of neurons in the human brain. These systems begin by processing information in a seemingly random fashion and progress towards improving the flow of data through learning, enhancing their performance and comprehension.

However, this intrinsically demands a lot of computing resources, leading to inefficiency and an inseparable relationship that requires the entire process to be restarted when adjustments are needed midway through. But the approach of selective forgetting revealed in this research allows for more efficient use of overall computing power by erasing key information from previous learning stages, reaffirming that forgetting is not a flaw but a function that enhances efficiency and adaptability.

In the digital age, our bodies are increasingly intertwined with technology, from wearable devices that monitor our health to smartphones that extend our cognitive capabilities. We see too many things through our smartphones. The opportunity to watch someone's life moments is called content, and we have gained the right and opportunity to record many things about individuals within digital platforms.

However, we are aware that the vast amount of personal information recorded in this way can become a burden, leaving behind indelible traces that we want to erase but cannot, prompting us to build new memory systems. This constant flow of information offers unprecedented access to knowledge but also presents challenges to our mental and physical well-being. Therefore, the concept of forgetting can be a crucial benchmark for managing this digital overload and prioritizing and processing the massive amounts of data we encounter daily.

From a neuroscience perspective, the general scientific view is that forgetting is a flaw in our memory system. For neuroscientists, forgetting was simply something to be fought against with all their might. However, from the perspective of neurobiology and computer science, forgetting is not only a normal process but also beneficial to our cognitive abilities, creativity, emotional well-being, and social health.

In his book "Forgetting and Freedom," philosopher Kang Shin-ju argues that forgetting should not be understood as a kind of simpleton state or simply a decline in memory. Rather, he argues that forgetting holds a meaning of active power to transcend and intense struggle for humans who want to stay in the past. From his perspective, forgetting is interpreted not as loss but as a necessary process for happiness, hope, and living in the present. As humans are self-centered beings to the point of obsession, they must go through a stage of emptiness to gain the opportunity to face others. Therefore, his assertion that forgetting can actually be another form of robust health reveals more than one might think.

As AI integrates into our daily lives, we will be reconsidering our relationship with our bodies in more diverse ways. The similarities between the efficiency of forgetting in AI systems and its benefits to human cognition represent another process in understanding the subtle ways technology reshapes our physical and mental landscapes. In the scientific field of creating AI, embracing the paradox of forgetting has led to the creation of revolutionary processes that were previously unseen. Then, how can we apply this paradoxical forgetting to our bodies, which have provided us with the concept, role, and understanding of forgetting? It seems like a question we need to ask.

References

Improving Language Plasticity via Pretraining with Active Forgetting

Selective Forgetting Can Help AI Learn Better

망각과 자유⁠⁠⁠⁠⁠⁠⁠