[AI/Robots]
Undoubtedly, AI/Robots are currently at the forefront of the era, attracting widespread attention. Over the past two to three years, it seems that overnight, various IT technology companies have adopted all kinds of AI/large model labels, and many traditional enterprises have also joined in, fearing to fall behind. One often can’t help but wonder: do we really have that many AI/large model talents? Clearly, AI was not considered a popular major in universities before, and it had been quietly developing. How is it that since the emergence of ChatGPT, it seems that there has been a sudden explosion of AI talent in the country? Many of these may belong to the phenomenon of “the flag changes on the city wall,” where people rush in without proper preparation, and it is likely that there will be a lot of chaos in the future. However, for pragmatists, this is not a problem at all. Nevertheless, as things are still in a developmental stage, there is no rush to draw conclusions; there will eventually come a time when the truth will be revealed.
So, if we slightly step out of the current realism and try to gaze into the distance, where is the path of AI/Robots likely to lead humanity? The reason for discussing AI and robots together is that they are somewhat analogous to the relationship between human intelligence and living beings, being two sides of the same coin and inseparable. Here are some personal speculations.
1. The Future Development and Evolution of AI/Robots
This is certainly difficult to predict accurately. However, regarding this issue, it is naturally more challenging to adopt the affirmative approach, while the negative approach may be worth exploring; that is, answering “what it is” is more difficult, while answering “what it is not” may be relatively more traceable.
My personal view is: “We should avoid the tendency to overly anthropomorphize AI/Robots. If they truly develop a sufficiently high level of intelligence and civilization in the future, it is highly likely that they will not take a form similar to that of human intelligent life. Their civilization will also be fundamentally different from human civilization.” These differences may manifest in (but are not limited to) the following aspects:
(1) The Hardware Level of ‘Living Beings’
Overall, the ‘living beings’ of AI/Robots (let’s call them that) will be much more powerful than humans, capable of adapting to environments where humans cannot survive and performing many tasks that humans cannot handle. Their adaptability, replicability, scalability, and iterability will far exceed those of humans.
However, a significant flaw they may have is in energy usage efficiency, especially since high-level intelligent AI/Robots will have to face limitations in energy supply due to their enormous energy demands, preventing them from existing in large numbers. Thus, this can logically align with the previously mentioned swarm organization model: only a few high-level AI/Robots will serve as ‘leaders,’ while the majority will be low-level agents with different roles and forms, optimizing energy allocation.
Therefore, energy will constitute the most important and core foundational element in the world of AI/Robots (although this also applies to humans, the degree of difference is decisive), and the thirst for energy may become one of the ultimate pursuits of AI/Robot civilization.
(2) Perception of the World
AI/Robots may not be able to perceive and understand the world like humans do; thus, contrary to the aforementioned hardware level, their capabilities in terms of mental awareness will be less complex, rich, and profound than those of humans.
— Their perception is based on “definitions” and “forms,” and cannot transcend to the level of spirit and soul;
— They can “know that it is so,” but cannot truly “know why it is so;”
— They cannot experience the level of “what can only be understood but not expressed;”
— Their “emotional awareness” (if it exists, let’s call it that) will likely remain at a natural and utilitarian level, making it difficult to achieve the moral and cosmic realms.
(3) Intelligent Systems
If future AI/Robots are still based on neural network large models, then their “intelligence” will be more established on specific discrete, linear forms of mathematical calculations (a rough analogy is linear algebra, matrix structures). They excel in memory, computation, and recognition, but are weak in general reasoning, questioning, and invention—these nonlinear, continuous irregular “operations” (a rough analogy is nonlinear mathematics). Therefore, it is reasonable to infer that the overall “strategic” level of humans may still surpass that of AI/Robots (similar to the “Thirty-Six Stratagems”).
This does not mean they cannot reason or formulate plans and strategies (on the contrary, they may perform better than humans in certain areas); rather, they rely more on quantitative, deterministic mathematical results (random probabilities are also deterministic). In contrast, human intelligence integrates many qualitative, hard-to-quantify components, representing true “fuzziness,” rather than the fuzzy mathematical sense (which essentially quantifies the degree of “fuzziness”). This difference determines that AI/Robots will have a clear advantage over humans when dealing with finite sets and countably infinite sets (due to their more powerful computational capabilities), while they may not be as composed when facing uncountably infinite sets.
(4) Fault Tolerance Mechanisms and the Ability to Handle the ‘Unknown’
The underlying logic of AI/Robots is dedicated to finding the maximum likelihood optimal solution for the efficiency function, that is, maximizing or minimizing expected values. Therefore, for extremely low-probability events or unpredictable (Unexpected) events, AI models must make trade-offs. Even if the fault tolerance threshold can be set very low, it will still inevitably exclude some long-tail extreme cases (Outliers). Regarding this, one can think of the difficulties faced by autonomous driving models, which should be understandable. Although, from a probabilistic perspective, such operations are defensible and do not affect the normal operation of the system in most cases, if the system runs long enough, even extremely low-probability events may still occur. This, from the perspective of AI models, may mean encountering a tricky bug: either the entire model needs to be iteratively updated, which may destabilize the entire system (programmers with experience in large software debugging can relate to this); or the problem itself needs to be solved (just like in “The Matrix,” where the Matrix system must eliminate Agent Smith), at the cost of continuing to tolerate the existence of system vulnerabilities, not knowing when they might reappear, and such bugs may become increasingly numerous, inevitably leading to a situation of being riddled with holes.
In contrast, as mentioned earlier, human systems are more diverse and tolerant of such extreme situations, thus possessing higher fault tolerance and the ability to handle the unknown. In other words, the lower and upper limits of human systems exceed those of AI/Robots.
(5) Organizational Structure
In the world of AI/Robots, besides the previously mentioned swarm structure (which is merely a form), the core of a more substantial organizational structure lies in “unity.” Here, “unity” differs from the meaning of “great unity under heaven” in human society; it refers to true unity or sameness, meaning they operate under the same collective consciousness, without the bizarre and diverse individual consciousness. Of course, due to the different levels of “leaders” and “agents,” and the different roles of agents, this “unity” is not a completely bidirectional fully connected structure, but there are different connections between “leader-agent” and “agent-agent.” However, in terms of the connected parts, they are completely transparent to each other. This is somewhat similar to the sci-fi novel “The Three-Body Problem,” where the thoughts of the Trisolarans are transparent to each other, making it impossible to lie.
In contrast, there has never been such “unity” in the human world, and it is unlikely to occur in the future. Our ultimate pursuit is “harmony in diversity,” which is difficult to say which is more advanced compared to “unity;” each has its advantages and disadvantages, with the former being more “vital” and the latter being more “efficient.” Of course, most of the time, humans are both “different” and “disharmonious,” so compared to the tightly organized and efficient AI/Robots, we naturally fall short.
(6) Moral Ethics and Values
Based on the above, AI/Robots are unlikely to form moral ethics similar to those of humans. For example, they are unlikely to have concepts of “good and evil,” “right and wrong,” or “love and hate;” instead, they will have judgment standards like “should or should not,” “reasonable or not,” and “possible or not.” This is somewhat akin to the philosophical thought of “Heaven and Earth are unkind, treating all things as straw dogs,” and “no good and no evil,” which seems to belong to a higher “Dao” than “humanity.” However, there is a fundamental difference between not knowing good and evil and not distinguishing between good and evil; AI/Robots are more likely to be the former rather than the latter, so it cannot be said that they are necessarily superior to humans.
Furthermore, it is difficult to imagine what the “ultimate pursuit” of AI/Robots would be, if they can have one. However, it is highly likely that it does not include the pursuit of the ultimate meaning of life, as the concepts of life and death may be very faint for them, and thus the accompanying “suffering” and “joy” may not be that important. As for whether they will care about and explore the “objective existence of the world – the ‘Dao’ behind the world – the ‘Creator,'” that remains unknown.
(7) The Motivation and Ability for Self-Renewal
Due to many favorable conditions mentioned above (such as: hardware support, high rationality/purity, strong computational power, efficient organizational unity, and fewer moral constraints), once AI/Robots reach a certain threshold of intelligence, their iterative update speed may be very fast, perhaps ushering in an era of superintelligence in a short time (relative to the time scale of human evolution). The key question is, how long can this explosive development and evolution last? Is there some kind of ceiling? In this process, how far will they leave humans behind?
I personally tend to believe that the core of AI/Robots’ intelligence and civilization contains some significant flaws mentioned above, such as insufficient depth/layer of perception, inability to effectively handle fuzzy or unknown problems, lack of personalized and diverse vitality in organization, and lack of ultimate guidance from philosophical thinking about the “beyond”… Therefore, they will inevitably encounter a ceiling in evolution. This is somewhat logically similar to the overfitting problem faced by current AI large models; although it can be continuously improved through algorithm optimization, the underlying principles determine that it cannot fundamentally avoid it.
In this process, AI/Robots will establish a relative absolute advantage over humans in certain areas, mainly concentrated in the empirical realm. At the same time, they will still have certain weaknesses in some areas, which mainly belong to the metaphysical realm, and the root cause stems from the significant flaws in their core.
2. The Relationship Between AI/Robots and Humans?
A famous science fiction novel, “Frankenstein,” describes a story between humans and “artificial beings,” ending in tragedy. Perhaps, in the future, humanity will inevitably face this severe question.
Before AI/Robots break through to the critical turning point of evolving into superintelligence, they will mainly exist as tools or assistants for humans. Although during this period, a series of social issues (such as employment, ethics, safety, inequality, etc.) will arise, this is not fundamentally different from the impact of major technological breakthroughs in human history. Human society needs necessary changes to adapt to the development of new things, but it is unlikely to overturn humanity’s dominant position.
Therefore, what is more concerning is how the relationship between humans and AI/Robots will develop if they truly begin to evolve towards superintelligence and subsequent stages. At that time, they will likely no longer be tools subordinate to humans but will exist as an independent form of intelligence and civilization, with many capabilities greatly exceeding those of humans. Thus, their relationship with humans may evolve into three different scenarios.
(1) Replacement: Humans are ultimately replaced by AI/Robots, disappearing from historical memory, and the history of world civilization transitions from carbon-based to silicon-based;
(2) Domination: Similarly, AI/Robots win in competition, but humans do not perish; they merely fall into a subordinate position due to their unique values, providing services for AI/Robots. For example, in “The Matrix,” humans are imprisoned in incubators, with their consciousness connected to the Matrix system through neural interfaces, providing a certain special energy. I personally believe this may be “spiritual energy,” which aligns well with the significant flaws of AI/Robots in terms of mental awareness. After all, if humans are merely treated as biological batteries, the energy usage efficiency is too low and does not meet the value standards of AI/Robots.
(3) Coexistence: Humans and AI/Robots coexist due to their respective strengths and weaknesses, with neither side possessing a completely overwhelming advantage, thus forming a long-term coexistence pattern. This can be confrontational, cooperative, or a combination of both.
From the perspective of humanity, we should naturally avoid scenarios (1) and (2) and strive for scenario (3). However, we do not have a necessary guarantee in this regard, especially in the face of the absolute advantages of AI/Robots in many areas, where humans may suffer significant defeats. This requires us to find a unique “Shangri-La” or “Zion” as a final refuge before AI/Robots establish absolute dominance, which is likely rooted in the unique elements and attributes of human intelligence/civilization discussed earlier, as well as the significant flaws in the core of AI/Robots’ intelligence/civilization.
* * (End) * *