In this blog post, we’ll take a chronological look at why artificial intelligence was created and how it has evolved, against the backdrop of the heightened interest that followed AlphaGo.
Early Objectives and Concepts: From Neuron Models to Thinking Machines
Interest in artificial intelligence surged following the match between AlphaGo and Lee Sedol, and many people came to view it as an “intelligence that assists humans.” But was the original purpose of artificial intelligence really just to help humans? To find the answer, we need to examine its early concepts.
The roots of artificial intelligence trace back to the early 1940s. Early researchers began with experimental attempts to understand how neurons function within the brain, aiming to create models that would simulate brain activity. Consequently, the focus of early AI research was on how to replicate the “brain” and its constituent parts—the neurons.
In the 1950s, the Turing Test, proposed by Alan Turing, established a criterion for determining whether a machine could “think” like a human, thereby sparking a lively debate about “thinking machines.” Subsequently, at the 1956 Dartmouth Conference, the term “Artificial Intelligence” was used for the first time, and AI began to establish itself as an academic field.
Growth, Setbacks, and a Resurgence: From Expert Systems to Machine Learning
In the 1950s and 1960s, scientists harbored expectations that machines could mimic human intellectual abilities, such as solving and proving mathematical problems. At one point, optimistic predictions were even made that machines with intelligence surpassing that of humans could be created. However, reality did not go smoothly.
Starting in the late 1960s, a series of failures ensued. Efforts to achieve practical goals, such as machine translation (a system that automatically translates one language into another), yielded results that fell short of expectations, and the connectionist approach—which sought to model behavioral and mental phenomena proposed in philosophy and psychology—could not be properly implemented with the technology available at the time. These successive failures led to a decline in research funding, ushering in a dark period known as the “AI Winter.”
By the 1980s, artificial intelligence made a comeback under the name “expert systems.” Systems emerged that automated tasks requiring specialized knowledge—such as medical diagnosis, mineral exploration, and specific scientific fields—to take over human work, and they actually achieved significant results in various fields. However, expert systems were specialized for specific domains, had a narrow scope of operation, and did not provide general-purpose intelligence.
To overcome this limitation, researchers began exploring ways for machines to learn on their own, leading to the full-scale growth of the field of machine learning. In the 1990s, artificial neural networks were revisited in the search for structures resembling human learning, and the subsequent development of “deep learning”—a term referring to deep neural network technology—paved the way for today’s achievements. The learning method used by AlphaGo is also a branch of this machine learning.
Shifting Goals and the Choices Ahead
In summary, the goal of AI development began with early research on neurons, shifted to “machines that perform tasks in place of humans,” and has since evolved toward creating “self-learning intelligence.” Over the past 70 years or so, this objective has continued to change depending on the circumstances and the level of technology.
These goals will inevitably continue to change in the future. Among the possible futures, one scenario envisions AI becoming a valuable partner that aids human progress and significantly increases social benefits. On the other hand, there is also discussion of a scenario in which strong AI (or superintelligence)—possessing cognitive abilities equal to or surpassing those of humans—emerges and comes into conflict with humanity.
It is unclear which scenario will become reality. However, what is important is to take an interest in these possibilities, closely observe trends, and actively consider how to prepare for the challenges we may face. By doing so, we will be better positioned to harness the potential of artificial intelligence for the benefit of human welfare and progress.
I hope that you, the readers, will also maintain a steady interest not only in the history of artificial intelligence but also in the issues likely to arise in the future, and actively participate in the discussion. The direction of technology ultimately depends on societal choices, and each of our interests and preparations will shape the future.
