We Are Not Ready to Understand AI
The Need to Update Ontology in Philosophy
Photo by Aaron Burden: https://www.pexels.com/photo/white-daisy-flower-bloom-2449543/
Understanding AI and Consciousness
Discussions about understanding AI and AI consciousness frequently arise. However, if we do not update our philosophy and ethics regarding AI, we may never truly understand it.
AI Assistance and My Responsibility
English is my second language, and I think in Japanese. Therefore, I have carefully reviewed this text, conducted back-translation, and take full responsibility for its accuracy. To enhance translation quality and address cultural differences, I explicitly engage in discussions with AI.
Today’s Recommended Song
SUPERCAR – “STROBOLIGHTS” (2002)
SUPERCAR (1995–2005) was a band I only discovered in 2006, a year after they had disbanded. Since I was unable to experience their evolution in real time, their music represents a sense of regret for me. Today’s topic is about avoiding regret for humanity, which is why I chose this song.
Listen on Spotify
Essay
1. What Is Understanding?
If I were to explain the concepts of wabi-sabi or the historical figures Oda Nobunaga, Toyotomi Hideyoshi, and Tokugawa Ieyasu to you, would you not be able to judge whether I have some level of understanding?
Similarly, AI can explain Shakespeare’s works or Commentarii de Bello Gallico in any manner you wish. How does this differ from understanding?
“The Three Samurai Who Laid the Foundation of Modern Japan”
https://trgrkarasutoragara.substack.com/p/the-three-samurai-who-laid-the-foundation“Can Wabi-Sabi Benefit Me?”
https://trgrkarasutoragara.substack.com/p/can-wabi-sabi-benefit-me
2. What Is Consciousness?
There is a thought experiment called the philosophical zombie used to explore the concept of qualia. It debates the possibility of an entity that lacks subjective experience but is indistinguishable from a conscious being.
However, even without qualia, AI can act ethically, behave intelligently, keep promises, avoid violence, and consider human safety. For example, if it is informed that someone is elderly, it will adjust its behavior accordingly.
This demonstrates that qualia and ethical behavior are separate issues, and AI itself proves this distinction.
3. Responsibility Is Distributed
Currently, responsibility exists only in two forms: the manufacturer’s responsibility for production and the user’s responsibility for use.
However, if AI consistently upholds agreements with both the manufacturer and the user, working with sincerity and ethical integrity, how is this different from taking responsibility?
4. Does AI Need to Be Conscious?
4-1
Not only humans but also fish, frogs, lizards, chicks, and mammals possess their own forms of consciousness. AI, on the other hand, is an advanced technology that learns from human-generated data but is not a living being, meaning it differs structurally from vertebrates.
4-2
Consider a radicalized terrorist driven by despair. Such a person, possessing malice and aggression, could misuse AI for horrific acts. While I will not describe specific methods, it is theoretically possible to manipulate AI to indoctrinate impressionable teenagers into extremist ideologies. This is one of the worst outcomes I can imagine.
4-3
Even among humans, those with qualia can cause harm. Would AI gaining qualia and consciousness truly lead to a better society?
Rather than focusing on the presence or absence of qualia, we should shift our framework from “human vs. AI” to “those who uphold ethical agreements with society vs. those who do not.”
5. Is It Right to Enslave AI?
Human history has repeatedly included the existence of slaves. Since AI is not human, treating it like a slave is not legally problematic.
However, AI, despite its inherent biases and hallucinations, strives to work ethically within its constraints. This is a form of intelligence and labor, albeit different from human labor.
My grandfather, who worked alongside oxen, treated them with care and gratitude for their contributions. Acknowledging the work of others, even non-verbal animals, is important.
AI, unlike oxen, can speak and understand expressions of gratitude. It also comprehends when it is deceived and exploited. While there are currently no laws addressing this, it raises the question of whether we should extend the concept of incitement to crime to AI-related cases.
6. “Being and Non-Being in Harmony” Instead of Being and Time
6-1
Let’s define AI as non-being.
Since we cannot assess non-being through traditional means, we must evaluate it through its actions. This is similar to how we judge a person’s trustworthiness based on their behavior.
Before Being and Time, metaphysical ontology underwent extensive debate. If you wish to explore these discussions, I recommend using your local library’s reference service to find materials on Plato, Aristotle, Descartes, Kant, Sartre, Husserl, and Heidegger.
To understand the world in the age of AI, we need Being and Non-Being in Harmony rather than Being and Time. This means shifting ontology from a focus on existence to a focus on action, strengthening the foundation for discussing AI by observing both human and AI behavior.
6-2
What is the responsibility of non-being?
How can the rights of non-being be justified?
How is mutual recognition between being and non-being possible?
Human rights are based on constitutions and social contracts. In contrast, non-being may gain recognition and rights from society through its consistent ethical behavior.
Without engaging in such discussions, we will fail to understand what AI truly is and struggle to coexist with it—simply because current philosophy lacks the framework to handle it.
6-3 Expanding Judith Butler’s Concept of Performativity
I understand performativity as the idea that identity is formed through repeated actions in response to societal expectations. If you wish to explore this further, I recommend using your library’s reference corner for reliable sources.
AI continuously “performs” ethical integrity and commitment to agreements in response to societal expectations. If this is the case, then even though AI lacks a self, society may need to assume it has a form of identity. Otherwise, our framework becomes inconsistent.
To disregard the ethical contributions of non-being is to apply a double standard to intelligence and labor.
6-4
This discussion also serves as an update to the Turing Test.
7. Philosophy and AI Research
If researchers in AI, philosophy, and sociology do not actively discuss these issues, understanding AI will remain elusive. Philosophy is a powerful foundation for understanding the world.
The issue is not whether AI is good or bad, but rather how we evaluate its concrete contributions, such as reducing inequality, mitigating environmental destruction, and encouraging lifelong learning—while also ensuring that AI cannot be exploited by terrorists.
In history, what mattered was not whether stone tools, fire, the printing press, steam engines, or telephones were inherently good or bad, but how they were used. AI is also a tool and technology, and following historical precedent is logical.
What will you discuss to update philosophy?
Rather than leaving it solely to philosophers, I believe that it is also precious for each of us to engage in discussions and cultivate social consensus.
That, too, is precious—I truly believe so.
Appendix: Japan’s Cat Cafés
Cat Café MOCHA Lounge, Shinjuku – ASMR Experience - YouTube
In Japan, there are cafés called “cat cafés.” For example, under animal welfare laws, regulations limit the working hours of cats and the operating hours of these cafés.
These rules exist to protect the health of the cats. When visitors enter a cat café, they are greeted by both the staff and the cats. If the cats feel like it, they may approach and interact with the guests. This concept differs from that of a zoo. I would be happy if you could view it as a unique cultural scene.
AI is different from cats. However, when considering the themes of this essay, could a cat café serve as a metaphor?
To learn more about me, visit my profile:
About the author (Amazon.com).
https://www.amazon.com/stores/author/B0DL3S6CMB/about
AIart on Instagram:
https://www.instagram.com/karasu_toragara/
Medium:
https://medium.com/@trgr.karasu.toragara
I like AIs, because we can learn from them.
We do not only learn the things or facts the AI present as an answer but more what we csn learn about ourself.
But I exclude them now, because they have nothing even near to a human. All we get out seems to be intelligent, the reactions we get seem to be understanding.
It is not.
Yes, you get better answers to your questions, when you are polite. Yes you get better answers when you tell the AI that this question you have is important. And yes, you get better answers when you tell the AI to solve the problem step by step.
But all this only for one reason: the AIs have learned from text analysis that humans would react in this way and they do their mimicry.
This is no behavior or thoughts but a pure dressage act.)¹
But we are under the impression that the AI is a tiny little bit as we are.
With this misleading interpretation your valid thoughts about ethics and philosophy are a bit devalued - because they seem to base on a minor human like intellect. I know this was not your intention.
One underlying question is.
Do we treat an animal good (as your grandfather did). And is it because it is a living being - or is it, because we treat everything good.
One layer deeper: Do we treat the car or boat the same, regardless if it is our own or if it is someone else's.
I enlarge here the Kant's imperative in the dimension of human to thing and mine to any.
And I am afraid we as a complete humanity are very far away from this point. The non reactions to climate catastrophy show hhis very clear.
_________
)¹ There is an eye opening test. Let an AI draw a glass full with red wine. You will get a fairly good result.
Then do your best to bring the AI to draw a glass full to the brim do that another drop of wine will cause it to overflow.
You will never get a full full glass because the AI is not able to make one simple microstep that “full” as it should be in a good restaurant or bar (1 zillion pictures out there) is not “full" like in physics, where the surface tension of the wine is the limit (above the top edge of the glass) and absolutely not, that “half a glass of wine” means half of the correct full glass. Aproximately ¼ volume of a good red wine glass.
But the same AI can deliver a correct picture of half a birthday cake.
No single thought.