A brand new evaluation argues that AI could by no means really suppose like people as a result of crucial elements of human intelligence can’t be programmed into machines.
A distinguished pc scientist argues {that a} proposal made by Alan Turing, extensively considered the daddy of theoretical pc science, despatched artificial intelligence research in the wrong direction for the past 75 years.
In his new analysis, “Turing’s Mistake: Escaping the Yoke of Unintelligent Machines,” Peter J. Denning examines ideas Turing advanced in 1950. At the time, many scientists believed that human intelligence could exist independently of the body and might therefore be recreated as software running on a digital computer.
Denning also disputes the idea that machine intelligence can be demonstrated through an imitation game (now known as the Turing test).
“These two claims have shaped much of AI research and development,” Denning writes. “My premise is that our acquiescence to these claims has led to the AI mess in which we find ourselves today.”
According to Denning, the artificial intelligence (AI) systems now being developed are unlikely to produce human-level intelligence, known as artificial general intelligence (AGI). Instead, he warns, they may create serious dangers without ever thinking like humans.
Why Tacit Knowledge Matters
Central to Denning’s argument is the idea of tacit knowledge. This refers to the enormous amount of human understanding that people possess but cannot fully express in words or translate into symbols that a machine can process.
Denning describes five broad forms of tacit knowledge that he says ‘elude machine learning’. They include common sense, everyday interactions with people and the environment, feelings and perceptions, practical skills, and the cultural and historical background shared by societies.
Researchers have spent decades trying to record common sense in a form computers can use. Beginning in the 1980s, Douglas Lenat’s ambitious Cyc project set out to build a vast database of common-sense facts. After 40 years of work, the project contained 25 million entries.
“Yet even this treasury could not add up to a background of common sense sufficient to make expert systems smart enough to be experts,” Denning notes. “Cyc validated that much of the knowledge that makes people experts cannot be articulated as propositions.”
Knowing What Is Not the Same as Knowing How
Practical skill creates another major obstacle, Denning argues.
“Our performance skills in thousands of domains cannot be communicated to machines,” Denning explains. “Whereas descriptions of skillful outcomes (‘know what’) can often be represented as bits and stored in a machine, we do not know how to encode the embodied knowledge for skillful performance (‘know how’).”
Music offers a clear example of this difference. Denning says: “A virtuoso violinist can play beautiful music yet cannot describe to an acolyte how to produce it.
“Even if a robot could observe and imitate skilled humans, having no biological body, a robot cannot grasp how the musician feels when playing beautiful music or how an audience feels when hearing it.”
Intuition, gut feelings, spontaneous creativity, and imagination are other forms of tacit knowledge that resist being reduced to computer instructions.
The Representation Problem
Denning calls the central obstacle ‘the representation problem’.
Computers can only perform calculations when data and instructions are encoded in physical forms they can recognize and process. Tacit knowledge, however, cannot easily be converted into such a format.
“Behind every word is a deep well of tacit knowledge that gives it meaning,” Denning says. “Words are but symbolic representations of meanings, not the meanings themselves. Commonly used Large Language Models, such as ChatGPT, Claude and Gemini only manipulate words, they cannot know or understand the meaning of what they are saying.”
This creates what Denning sees as an unbridgeable gap. Because scientists do not fully understand how tacit knowledge operates within humans, they cannot determine how to transfer it to a machine.
“How we host tacit knowledge is largely a mystery,” Denning admits. “All we know is that it is embodied. We have no idea what we might observe and measure in our bodies to reveal it.”
Why Context Changes Meaning
Denning also stresses the importance of context, or the surrounding circumstances that give human words and actions their meaning and purpose.
A statement can mean very different things depending on whether the speaker is sincere, sarcastic, angry, playful, or teasing. Context also helps people decide when to use humor, when to show tact, and how to interpret what someone leaves unsaid.
“When you inquire into where an assumption of the current context came from, you discover it rests on previous conversations from previous contexts. Each of those in turn rests on further previous conversations and their contexts. This pattern is endless and fractal,” Denning explains.
Culture May Be Beyond Large Language Models
Culture presents a related challenge. It includes values, social norms, judgments, histories, communities, moods, and relationships involving power or care.
“Human conversations are imbued with background assumptions that give meaning and relevance to the words being used,” Denning explains.
He argues that making large language models larger will not solve this problem.
“Scaling up LLMs with ever larger neural networks will not enable them to acquire the embodied human knowledge we call culture. LLMs will not attain the objective of the Turing test: to demonstrate machine thought indistinguishable from human thought.”
Denning ultimately describes a form of mutual incomprehension between people and machines. Artificial neural networks may develop their own kind of machine tacit knowledge, but humans may be unable to understand it.
“Machines cannot read our tacit knowledge and we cannot read theirs,” he writes. “We are aliens across an uncrossable divide.”
The AI Safety Risk
This divide could have major consequences for AI safety. If machines cannot understand the unstated context behind human instructions, Denning warns that reliably aligning their behavior with human goals may be impossible.
“Through AI automation, agentic networks of machines are likely to develop their own machine intelligence that does not reach the level of human general intelligence but is still quite capable of creating severe problems for humans. This threat is a greater than a take-over by superintelligent machines,” he explains.
In Denning’s view, the most immediate danger is not a superintelligent machine that surpasses humanity. It is a network of less intelligent systems that acts in powerful, unpredictable, and potentially harmful ways.
“Machine intelligence has different concerns from us and does not appear to care about us. Its ways of thinking and problem-solving look alien to us. We do not yet know how to live safely with these machines.
“Pulling back from an AI automation singularity will demand much from us. We start by accepting that the familiar culture is fading away as intelligent machines appear in our society and we do not know what is coming. We decline to think like machines or be subservient to machines. We refuse to submit to a yoke imposed by low-intelligence machines. Most importantly, we reassert our humanity, declare once again what makes us different from machines, and celebrate those differences.”
Never miss a breakthrough: Join the SciTechDaily newsletter.
Follow us on Google and Google News.
