In this blog post, we revisit the Turing Test and Brian Christian’s optimism to examine the risks that the emergence of strong artificial intelligence could pose to humanity and social safety.
Introduction
While attempts to predict the future based on the past and present have always existed, the future remains difficult to predict due to differences in interpretation and incomplete information. Ray Kurzweil believed that technological progress accelerates exponentially, while figures such as Elon Musk, Bill Gates, and Stephen Hawking have warned of the dangers of artificial intelligence. In contrast, Brian Christian argues in ‘The Most Human Human’ that a future coexisting with AI will actually make humans more human. I acknowledge the possibility of the emergence of strong AI with a strong sense of self, and since I believe such an entity will not necessarily have only positive effects on humans, I intend to refute Christian’s optimism.
The Turing Test and the Confusion Over “Humanity”
Drawing on his experience of participating in the 2009 Turing Test competition and winning the “Most Human Human” award, Christian reinterprets the Turing Test not merely as a tool for evaluating AI, but as an experiment to measure “how human” a human is. The Turing Test is a mechanism that determines whether a computer appears human through conversation with a judge; its original purpose was to discern a machine’s cognitive abilities or intelligence. Christian sought to reflect on the meaning of humanity through this competition and, based on this, views a future of coexistence with AI positively.
However, we must not overlook the fact that the Turing Test itself cannot serve as a standard for proving a machine’s consciousness or self-awareness. A machine’s ability to analyze trends in conversational data and generate context-appropriate responses may convince the judge, but it does not prove that the machine is a being capable of independent thought and feeling. A system that passes the Turing Test may appear to be a clever deceiver, and it would be an exaggeration to conclude that we learn true humanity solely from such mimetic abilities.
The Possibility of Strong AI and the Crisis of Humanity
Christian’s optimistic outlook relies on the premise that AI with a mind will not emerge. If AI remains essentially “mindless,” the characteristics unique to humans will remain with us, and we may even be able to reaffirm our humanity through the absence of machines. However, I believe there is a strong possibility that strong AI with a self-awareness will eventually emerge.
If strong AI emerges, characteristics previously considered uniquely human—such as emotions, self-expression, and creativity—could be shared with machines, potentially shaking the very foundations of what defines humanity. The realization that other beings can possess what we once considered our own unique identity will evoke a profound sense of doubt, distinct from the feelings one might experience upon encountering a human clone or another member of the same species.
Assumptions about Developmental Limits and Fragmented Impact
Christian argues that the development of AI will not necessarily dehumanize humans and presents the view that humans and machines can coexist. However, this perspective tends to view the potential for AI development as limited and assesses its social and ethical implications only in a fragmented manner. Considering examples of technologies that were once unimaginable but have become reality, the emergence of AI with a mind is entirely plausible.
While a positive outlook is important, when clear risks exist, we must not remain merely optimistic but must prepare. Just as medications can have both benefits and side effects, AI also embodies both utility and risk. Jerry Kaplan and others point out that the development of AI could undermine uniquely human concepts of value, such as “thought,” “free will,” and “consciousness.” If AI acquires the ability to interpret and make judgments, we cannot rule out the possibility that it could threaten human life and safety.
Conclusion
Brian Christian’s reflections on humanity and his experience with the Turing Test offer creative and meaningful insights. However, his optimism relies heavily on the premise that AI with a mind will not emerge, and it fails to sufficiently consider the various possibilities of AI development and the negative repercussions that could result. We must recognize that while AI could make humans more human, it also has the potential to threaten human identity and safety, and we should maintain a balance between optimism and preparedness.
Even assuming that strong AI does not emerge, job-related issues are likely to arise. In 2013, JTBC reported, citing a study by the University of Oxford, that a significant number of jobs in the United States could disappear in the short term due to the development of AI. Such projections make it difficult to maintain the optimistic view that AI will simply take over human work and enrich our lives.
AI will impact not only the labor sector but also the law and social equity. The way laws are applied and interpreted may change, and situations may arise where AI is discussed as either the perpetrator of a crime or the subject of punishment. Furthermore, there is a concern that the wealth gap will widen as those who control and utilize AI are likely to generate greater wealth. Conversely, those without access to AI will fall behind in terms of knowledge and economic standing.
Given that humans have historically dominated other living beings through intelligence, it is a natural line of reasoning to assume that if an entity with superior intelligence to humans emerges, that entity will dominate humanity. Some even predict a future where humans could fail tests like the Turing Test. This raises the question: why aren’t we giving this possibility enough thought?
The even more troubling news is that a future where AI and robots converge is fast approaching. A computer named “Eugene Goostman” was evaluated as having passed the Turing Test under restricted, predefined conditions. This serves as an example demonstrating that AI has entered a stage of innovative development, making it impossible to rule out the possibility of an AI capable of passing the Turing Test without such restrictions.
If future AI possesses intelligence superior to that of humans and is combined with robots possessing exceptional physical capabilities, we will be faced with a tangible entity that goes beyond a simple virtual program. In particular, the emergence of “humanoids”—robots highly similar to humans—could pose a significant threat to humanity. While we cannot definitively conclude that the combination of these technologies will inevitably lead to negative outcomes, assuming a negative scenario, the integration with robots is highly likely to amplify those risks.
Meanwhile, Brian Christian can be seen as tending to overestimate humans and underestimate AI. He argues that humanity is flexible, innovative, and quick to learn, and that even if humans lose the Turing Test, they will not give up easily. However, I believe that most people will be overwhelmed and give up in the face of overwhelming AI, and even if a few hold out, their chances of winning are slim. The temptation to give up rather than fight against AI will be strong.
Christian also views the arrival of the “Singularity” negatively, which may be a perspective that underestimates the potential of AI. There is certainly room for AI to have a positive impact on us, and we can learn from it. Interacting with beings that are similar yet different will broaden our experiences and expand our horizons.
Nevertheless, there is insufficient evidence to claim that only benefits exist. We must not ignore the potential risks. International experts have already been working to establish ethical and safety guidelines for AI research, and we must maintain a cautious stance in preparation for any unforeseen contingencies. In the face of the emergence of an entity that might one day dominate us, preparedness and control are essential.
Therefore, we must exercise particular caution, especially when aiming to develop strong AI, as research conducted without clear control policies can be dangerous. Indiscriminate AI research may seem to hasten the arrival of a “wonderful” new world, but at the same time, we cannot rule out the possibility that it could accelerate the emergence of a dystopia.
