An infant interacts with a toy robot. Photo: TUCHONG

In recent years, the development and application of large language models have led to artificial intelligence (AI) becoming increasingly involved in areas such as human-machine interactions and content generation. As a result, today’s AI differs significantly from the traditional concept of AI. Traditionally, “artificial intelligence” refers primarily to “logical intelligence.” However, the strong interactive and generative capabilities of current AI have, to some extent, exceeded the scale of human-like or simulated intelligence in a “logical” sense, and may evolve into a form of intelligence on par with human intelligence. Given this shift in scope and capacity, this article proposes substituting “artificial intelligence (人工智能)” with “digital intelligence (数字智能)” when referring to current AI. 

Misleading name

“人工智能” derives its name from a direct translation of the English term “artificial intelligence.” In The New Oxford Dictionary of English (2001), “artificial” has two basic meanings. The first is its literal meaning: man-made; humanly contrived. The second is a metaphorical meaning: (with respect to people, behavior, attitudes, etc.) insincere; false; affected. Similarly, The Chinese-English Dictionary (2007) provides two basic definitions for “artificial.” The first definition encompasses three layers: man-made, false, and non-native. The second basic meaning encompasses two layers: man-made imitations, such as artificial flowers or artificial bait, and synthetic analogues, such as artificial fertilizers or other chemicals. The Chinese-English Dictionary also includes the term “artificial intelligence,” translated as: artificial intelligence (人工智能). 

When we combine the interpretations from both dictionaries, the basic meaning of “artificial” is “man-made” or “humanly contrived,” while its metaphorical meaning implies “false” or “fake.” According to semantic theories, the semantic value of “artificial” is negative, inherently associated with negative connotations such as “false” and “fake.” From the perspective of semantic connotation, “artificial” stands in opposition to “natural,” with natural things typically regarded as real, while artificial things are seen as false. Based on this, the term “artificial intelligence” may cause people to associate it with “fake intelligence,” meaning that no matter how powerful it becomes, it is ultimately a form of “fake intelligence” that is difficult to be on par with human intelligence. 

Another implicit meaning of “artificial” is the notion that humans maintain absolute control over anything they create. Based on this understanding, people tend to think of “artificial intelligence” as merely a tool or toy for humans. The fallacy of this view lies in confusing the two distinct types of artificial objects: one is mechanically and aesthetically artificial, serving as a tool or toy; the other is created through discovery, possessing a certain degree of subjectivity, and can even be regarded as a “non-human” agent. 

Furthermore, the term “artificial” carries several negative metaphorical connotations in regard to people, behaviors, and attitudes, such as insincere and fake. These negative metaphorical connotations may mislead people’s understanding of AI’s role in society and even its potential impact on humanity’s future. This may cause people to dismiss AI’s prominence as nothing more than “hype.” While there may indeed be many instances of hype in discussions about AI, the actual advancements and their impacts are not to be ignored. This misconception not only distorts the theoretical essence of generative digital intelligence and other similar concepts, but also risks misleading the public, preventing them from fully recognizing the opportunities and challenges that generative digital intelligence may bring to humanity’s future.

Logical intelligence 

A dictionary is not strictly an academic research product in essence, and its treatment of word meanings can only serve as a reference. To more comprehensively understand the meaning of a word, one should rely on relevant academic research findings as a theoretical basis. In fact, the meaning of “artificial” is much more complex than the definitions listed in The New Oxford Dictionary of English and The Chinese-English Dictionary imply. As early as 1910, American pragmatist philosopher John Dewey pointed out that “artificial” has both positive and negative connotations: the positive connotation means “professionally trained,” i.e. “logical”; while the negative connotation means “humanly contrived” and “false.” 

Since the word “artificial” has both positive and negative connotations, which applies in the context of “artificial intelligence”? In a strict sense, the positive connotation of “artificial” —“logical” —is related to mathematics or formal logic. Given that the underlying logic of generative digital AI is deep learning algorithms, and the foundation of algorithms is mathematics and formal logic, the true meaning of “artificial” in scientific literature on “artificial intelligence” is actually the positive connotation of “logical.” Thus, the true meaning of “artificial intelligence” is “logical intelligence.” However, technological advancements have, to some extent, diminished the negative connotations of “artificial,” leading to a certain convergence or unification between the “artificial” and “logical” aspects in terms of usage habits and actual operations in the field of AI. 

Currently, general dictionaries provide a rather broad definition of “artificial intelligence,” often failing to explicitly identify it as essentially a “logical intelligence.” For instance, the 7th edition of The Modern Chinese Dictionary (2016) defines AI as “a branch of computer science that utilizes computers to simulate human intellectual activities.” Similarly, the Collins English Dictionary, which selected “AI” as the 2023 Word of the Year based on its analysis of an 18-billion-word database, defines it as “the modeling of human mental functions by computer programs.” Through semantic analysis, these two dictionaries essentially define AI as a type of human-like or simulated intelligence. In reality, traditional AI could indeed be considered a type of “logical intelligence.” However, the powerful interactive and generative capabilities of current, new-generation AI have taken it beyond the realm of “logical” human-like or simulated intelligence, evolving into an emerging form of “digital intelligence” that competes with human intelligence. 

Digital intelligence 

At present, AI has taken on new forms that surpass traditional models, and given the potential confusion caused by the use of naming conventions such as “generative artificial intelligence” that continue to use the traditional term “artificial intelligence,” it may be more appropriate to refer to “artificial intelligence” at this stage as “digital intelligence.” 

Although contemporary AI is still a human creation, it has lost its mechanical essence as an artificial creation and has evolved into a new form of “digital intelligence” based on a “digital brain.” The powerful generative capabilities of large language models mainly stem from deep learning algorithms and massive amounts of data. Computers use pre-trained data sets as inputs to construct a “digital brain” using deep learning algorithms. Since the emergence of large language models, AI systems have modeled human behaviors and psychological processes through deep learning algorithms, surpassing normal individuals in terms of general intelligence. Considering the exponential development of AI over the past decade, the next 10 years may see such exponential iterations that many people could become “transparent” in the algorithmic sense. 

Humans may never fully control the emerging “digital intelligence” of generative digital technology. Like language itself, large language models are human creations, but the underlying logic behind them is not invented by humans. Rather, it is based on deep learning algorithms, similar to evolutionary principles. As some industry insiders believe, AI may no longer be an invention but more of a discovery, meaning humans cannot truly master it. Furthermore, intelligent systems may exhibit emergent properties, potentially evolving into “super-conscious digital entities” that may invert the power dynamic between humans and machines. 

At present, the agency of generative AI has to some extent already gained recognition by the science community. For example, ChatGPT was named among the top 10 scientific figures of 2023 by Nature. Currently, generative digital intelligence is evolving toward digital subjective intelligence. ChatGPT, for instance, adopts a question-and-answer format to present its generated results. Digital subjective intelligence is more complex, as it can function autonomously and engage in coordinated interactions to achieve set goals, allowing for collaborative decision making and actions that could lead to higher levels of intelligence. Idealized digital subjects are the same as human subjects, not only possessing the ability to plan, reason, correct errors, and reflect, but also exhibiting internalized “personality” traits, including human-like emotions or empathy. In addition, the human-computer interaction capabilities of digital subjective intelligence may be more worthy of attention. A simpler form of interaction is the “instruction-execution” method, which ChatGPT currently employs. A more advanced method would be a partner-type interaction, in which human and digital entities engage in multi-dimensional, in-depth collaboration as equals. Both the academic and business communities are actively exploring this more sophisticated form of interaction. 

In summary, generative digital intelligence and digital subjective intelligence are no longer traditional forms of “artificial intelligence,” but rather an emerging type of “digital intelligence.” Therefore, “digital intelligence” may be a more fitting term to refer to the current stage of “artificial intelligence.” 

Wang Fufang is a professor from the Institute of Linguistics and Foreign Languages at Beijing Foreign Studies University.

Edited by ZHAO YUAN