The Turing Mirage: Has 75 Years of AI Research Been Built on a Fundamental Error?

For three-quarters of a century, the roadmap for artificial intelligence has been guided by the North Star of Alan Turing’s 1950 vision. The objective has remained stubbornly consistent: to create a machine capable of mimicking human cognition so perfectly that it becomes indistinguishable from a person. However, a provocative new critique from Peter J. Denning, a luminary in computer science and a former president of the Association for Computing Machinery (ACM), suggests that the industry may have been chasing a phantom.

In his latest book, Turing’s Mistake: Escaping the Yoke of Unintelligent Machines, Denning posits that modern AI is currently mired in a "mess" precisely because it has spent decades attempting to solve the wrong problems. By tethering progress to the Turing Test and the assumption that intelligence is purely a software-based phenomenon, Denning argues that researchers have ignored the essential, embodied nature of human knowledge, leading us toward a technological precipice that is as dangerous as it is misunderstood.

The Foundations of a False Path

To understand the current state of artificial intelligence, one must return to 1950, when Alan Turing published his seminal paper, "Computing Machinery and Intelligence." Turing proposed that intelligence could be decoupled from biology—that it was, in essence, a computational process that could exist independently of a physical body. This assumption led to the second pillar of his theory: the "Imitation Game," or the Turing Test. If a machine could converse with a human such that the human could not tell whether they were speaking to another person or a computer, the machine could be said to be "thinking."

Denning argues that these two claims—the portability of intelligence into software and the validity of imitation as a metric for thought—have acted as a conceptual cage. "My premise is that our acquiescence to these claims has led to the AI mess in which we find ourselves today," Denning writes. He suggests that by focusing on Artificial General Intelligence (AGI), the industry has neglected the reality that human intelligence is not merely a collection of data points, but a deeply embedded, biological, and cultural experience.

A Chronology of Computational Hubris

The history of AI is often told as a triumphant march toward human-level capability, but viewed through the lens of Denning’s critique, it looks more like a series of failed attempts to bypass the "representation problem."

• 1950s–1970s (The Era of Symbolic AI): Researchers operated on the belief that human intelligence was essentially logical and rule-based. If we could just encode the right "if-then" statements, the machine would reason like a human.
• 1980s (The Knowledge Engineering Boom): Projects like Douglas Lenat’s Cyc were launched to manually input "common sense" into computer systems. After four decades and roughly 25 million entries, the project stalled. It became clear that common sense could not be stored in a database because it is not merely a list of facts; it is a context-dependent background that humans navigate effortlessly.
• 2010s–Present (The Generative AI Revolution): The shift to Large Language Models (LLMs) represents the ultimate doubling-down on Turing’s imitation premise. These models predict the next token in a sequence with uncanny accuracy, creating the illusion of understanding. Yet, Denning argues, they are doing exactly what they were designed to do: manipulating symbols without ever grasping the meaning behind them.

The Tacit Knowledge Problem: What Data Cannot Capture

The core of Denning’s argument rests on the concept of "tacit knowledge"—the vast, silent repository of human understanding that defies codification. Tacit knowledge is what allows us to navigate the world without constantly thinking about the rules of gravity, social etiquette, or the emotional state of those around us.

Denning identifies five distinct categories of tacit knowledge that remain impervious to machine learning:

1. Common Sense: The background awareness of how the world functions.
2. Everyday Interactions: The fluid, unpredictable nature of human social exchange.
3. Emotions and Perception: The sensory and visceral experience of being alive.
4. Practical Performance Skills: The "know-how" that separates a virtuoso from a novice.
5. Social and Historical Context: The layers of cultural, political, and personal history that imbue human communication with subtext.

"Whereas descriptions of skillful outcomes—the ‘know what’—can often be represented as bits and stored in a machine," Denning explains, "we do not know how to encode the embodied knowledge for skillful performance, the ‘know how.’" A violinist may be able to transcribe the notes of a concerto into a digital file, but the machine cannot replicate the visceral, physical connection between the performer, the instrument, and the audience’s emotional response.

Why Knowledge Resists Encoding: The Representation Barrier

The technical limitation, according to Denning, is the "representation problem." Computers function on binary inputs—data that has been stripped of ambiguity and converted into a physical format the processor can recognize. However, human language and thought are essentially "fractal" in nature.

"Behind every word is a deep well of tacit knowledge that gives it meaning," Denning asserts. "Words are but symbolic representations of meanings, not the meanings themselves." When a user inputs a query into a tool like ChatGPT or Gemini, the machine is not accessing a database of "understanding." It is executing a statistical simulation of linguistic patterns. Because the machine lacks a body, it lacks a stake in the world. It cannot experience pain, joy, or the weight of history, and therefore, it cannot truly "know" what it is saying.

Implications for Safety: The Alien Divide

Perhaps the most harrowing aspect of Denning’s thesis is his assessment of AI safety. The industry is currently preoccupied with the fear of "superintelligence"—a machine so smart it decides to replace or eradicate humanity. Denning finds this fear misplaced. The real danger, he argues, is not that machines will become too much like us, but that they will remain fundamentally "alien."

Because human and machine intelligence operate on different planes—one embodied and cultural, the other statistical and symbolic—they are separated by an "uncrossable divide." If we rely on these systems to make critical decisions, we are trusting an entity that cannot comprehend the tacit, unspoken context of our goals.

"Machine intelligence has different concerns from us and does not appear to care about us," Denning warns. "Its ways of thinking and problem-solving look alien to us. We do not yet know how to live safely with these machines."

The Path Forward: Reasserting Humanity

The implications for policymakers and the tech industry are profound. If AGI is an unreachable, perhaps even ill-conceived goal, the pursuit of it may be causing more harm than good by automating essential human functions that require judgment and empathy.

Denning calls for a radical reorientation of our technological goals. He suggests we must:

• Decline to think like machines: Avoid the temptation to reduce human complex social and cultural problems into purely algorithmic, data-driven solutions.
• Reject subservience: Refuse to allow the "yoke" of low-intelligence, high-automation systems to dictate the terms of our society.
• Celebrate the distinction: Recognize that what makes us "inefficient" compared to a computer—our intuition, our gut feelings, our messy cultural norms—is precisely what makes us human.

As we stand at this crossroads, the challenge is no longer about building a machine that can pass the Turing Test. It is about understanding the inherent limitations of that test and, in doing so, reclaiming the unique value of human thought. The future of technology should not be an attempt to replace human intelligence, but a conscious effort to ensure that our tools remain tools—subordinate to the values, cultures, and embodied wisdom that machines, no matter how advanced, can never truly possess.