Interesting debate, Mohamed el Nawawy

Whether machines can "think" is still a philosophical question that I personally enjoy in fiction and literature, although it seems like something real that could happen at some point in the future. But nowadays, AI systems like LLMs or smart warehouse systems are framed and utilized for recognizing patterns, making decisions, and optimizing operations. That doesn’t mean they think like us, because they lack two fundamental things from my point of view: emotions and intentionality.

For example, intelligent warehouse systems (my current topic) can track inventory or control robots using digital twins. It looks smart, but if something unexpected happens—like a broken forklift—they may fail without human help. It’s not exactly an “AI hallucination” like in LLMs, but it shows a similar kind of limitation: they don't know what to do next due to a lack of real understanding. Of course, this can be dealt with through human intervention—like writing a command that triggers emergency procedures or similar—but that’s exactly the point: human intervention is still needed because of AI's limitations.

They use models like neural networks or reinforcement learning, which are powerful but still limited—restricted by the human logic that created the mathematical formulas applied in AI computing. So, while AI can look intelligent, it lacks the awareness or flexibility we associate with true thinking.

Thanks for the debate.

Kind regards, Raúl Fernández

Dear Raúl Eduardo Fernández Pachas

Thank you for this interesting discussion

Kind regards

Mohamed

Dear Mohamed el Nawawy

If I may slightly reframe the question—which is both fascinating and deeply relevant in our present moment—perhaps we should not only ask whether machines can think, but rather what it would take for us to no longer recognize the difference between genuine thinking and its mere imitation.

This perspective echoes Alan Turing’s famous 1950 article Computing Machinery and Intelligence. Turing observed that the question “Can machines think?” is too ambiguous to produce a clear answer— at least in his time. Instead, he suggested a more pragmatic approach, what we now call the Turing Test: if a machine can engage in conversation so convincingly that a human cannot reliably distinguish it from another human, then, for all practical purposes, we may as well treat the machine as intelligent.

Seen this way, the case of an AI system that simulates linguistic understanding yet occasionally hallucinates responses—whether as a general-purpose chatbot or a specialized medical assistant—does not demonstrate the full reality of thought. Rather, it exposes the thresholds and limits of imitation. These examples remind us that the deeper philosophical challenge may not be to define once and for all what “thinking” truly is, but to identify the moment when simulation becomes so effective that the distinction between thinking and appearing to think no longer holds decisive weight.

Turing, Alan M. "Computing machinery and intelligence (1950)." Mind 59.236 (2021): 33-60.

Dear Marco di Stasio

Thank you for highlighting this challenging yet interesting perspective

Kind Regards,

Mohamed

Dear @Mohamed, I’m more aligned with Marco’s view. AI, machines, and LLMs are remarkably good at discovering and exploiting patterns and similarities. We are still in the “attention is all you need” era, where the dominant paradigm is next-token prediction—powerful in practice, yet fundamentally a mimicry of linguistic distribution rather than an emergence of genuine thought.

As systems improve at this mimicry, the real challenge will not be whether they think, but how we distinguish between authentic reasoning and highly refined statistical imitation.

Dear Christos Kalyvas-kasopatidis

I’d just like to highlight two points relevant to this discussion:

• Being unable to distinguish between authentic reasoning and highly refined statistical imitation for a system is not an indication of its genuineness. An example is deepfake where a system generates new content that seems to be original although it’s fake.
• The same concept applies to human attributes. A machine appearing to be conscious isn’t necessarily an indication of its consciousness.