# Why the Turing AI test is no longer relevant **By:** Denis O. **Published:** 2026-01-16T19:41:47.515Z **Source:** [LinkedIn](https://www.linkedin.com/posts/denis-o-b61a379a_ai-activity-7418014628711923712-KGEx?rcm=ACoAAAfsct4BVBNkOVI5cyYn-UOIIe2RSRhT-cI) --- # Denis O.’s Post [Denis O.](https://www.linkedin.com/in/denis-o-b61a379a?trk=public_post_feed-actor-name) Why the Turing AI test is no longer relevant Hear me out. Just bear with me for a sec. The Turing test made sense in its time. It was born in an era when computers were rare, mysterious, and fundamentally alien to everyday human reasoning. Alan Turing was not asking whether a machine could think in any deep or structural sense. He was asking whether a machine could convincingly participate in a human conversation without giving itself away. That framing mattered in the mid 20th century, when language itself was mistakenly treated as a proxy for intelligence (hmmmmm), and when interaction bandwidth between humans and machines was narrow, slow, and symbolic. If a system could hold a conversation, respond plausibly, and avoid obvious errors, that alone felt extraordinary. But the test measures something very specific, and very limited. It measures narrative plausibility under constrained interaction. It rewards surface coherence, social mimicry, and the ability to stay within conversational expectations. Plausibility and confidence get you surprisingly far. Bullshit gets you even further. Fooling humans is not nearly as complicated as the mythology around the Turing test suggests. Humans are cooperative by default. We assume good faith. We project intention. We fill in gaps. We want to believe other primates who sound fluent and confident (at least I do). Given average cognitive load, time pressure, and social norms, sounding reasonable is often good enough. That is precisely the problem. Passing for human is not intelligence. It is theater. It is narrative control. It is statistical fluency over symbols that already exist. You can sound right while being wrong. You can be persuasive without being grounded. You can generate endless text without owning any process that must survive contact with reality. None of that tells us whether a system can actually operate. The real test of AI is not whether it can talk its way through a chat. The real test is whether it can run autonomously under uncertainty. Whether it can maintain coherence while the world changes. Whether it can adapt without a human in the loop continuously steering, correcting, restarting, or approving every step. Whether it can absorb surprise, revise its internal state, respect constraints, and keep functioning over time without collapsing into noise or drift. Intelligence shows up when a system is embedded in a process, not a conversation. When failure has consequences. When inputs are incomplete. When conditions are non stationary. When there is no script, no prompt, and no safety net of human oversight. Adaptive autonomy is the bar. Continuous operation is the bar. Coherence under uncertainty is the bar.. Basically DARPA Wave 3 AI. The Turing test asks whether a machine can pretend to be human.. The future of AI asks whether a machine can be responsible for something real and mission critical without human in the loop. [#ai](https://www.linkedin.com/signup/cold-join?session_redirect=https%3A%2F%2Fwww.linkedin.com%2Ffeed%2Fhashtag%2Fai&trk=public_post-text) The Turing Test has never been about determining anything substantially meaningful regarding the capacity of a computer system but rather has always been an unscientific referendum on our omnipresent human susceptibility to being deceived. The test functions more like a Rorschach test in which the outcome is primarily indicative of the human evaluator's psychology and provides only notionally significant information about the capabilities or quality of the machine. Thank you, Dennis O. So does this mean the real test of AI isn’t whether it can act on its own, but whether it can keep making responsible, value-aligned decisions over time—without losing coherence or causing harm? I agree with the first part of the thesis (not least since you sound fluent and confident) However, isn’t it a too high of a bar? How many humans are intelligent according to these expectations? How many can indeed run autonomously (and effectively) under uncertainty? I suspect that many may find it difficult even under full certainty. The degree and type of uncertainty matters, mission criticality matters, type of systems and domain matter. Humans in most social systems are often still steered, corrected, seek approvals, follow specifically designed protocols and policies as well as unintentionally or deliberately break them, as well as step over constraints. On the other hand, humans have successfully designed systems that could reach space and land back fully unmanned. Those systems have adaptive autonomy, continuous operation and necessary degree of coherence under expected uncertainty, but we don’t call those systems intelligent. Operating under uncertainty goes beyond learning and requires innovation (creativity, innovation, problem solving). You can't go read a book or ask advice when, e.g., you have seconds to choose between life and death. Learning compiles while innovation fixes, answers, solves, creates. And reasoning can't function without innovation; these two are inextricably linked. This missing understanding is, and has always been, the core issue with the entire AI/AGI paradigm. Mimicry can't innovate. And yes, humans aren't good at innovation. They've spent decades trying to learn their way to it. Instead of the Turing test, the new goalpost for AGI could be the "Piaget Test". I'm a father to a toddler, and he amazes me at least once every week with how quickly he understands and adapts to the world. His memory and social mimicry are only a subset of his intelligence. If I were in AI research now, I'd get off the language model bandwagon and instead train a model through Jean Piaget's 4-stages of cognitive development: [Denis O.](https://www.linkedin.com/in/denis-o-b61a379a?trk=public_post_comment-text) I was about to write a post about this, and then I saw yours. Infinity shows that even the most rigorous systems have limits, and this connects directly to today’s AI debate. Cantor revealed that infinities come in different sizes. Gödel proved that any reasoning system will contain truths it cannot prove. Cohen showed that some questions are undecidable, depending on the universe you choose. Together, these results show that no system can ever be complete, perfect, or all‑knowing ;)