<![CDATA[FRANCESCO TAVA

AI can perform tasks that resemble critical thinking—such as analysing information, evaluating arguments, and identifying patterns or inconsistencies—but it does not truly think in the human sense. Unlike humans, AI lacks consciousness, self-awareness, understanding, and lived experience; it generates responses based on learned data patterns rather than genuine reasoning or reflection. While it can be highly effective in structured or data-driven contexts, its “thinking” is best understood as a simulation of reasoning rather than real critical thought grounded in awareness and judgment.

The paragraph above was generated by ChatGPT (GPT-5.3, 2026) in response to two simple prompts: “Is AI able to perform critical thinking?” and “Summarise everything into one paragraph”.

If this were a human response, we might call it honest – perhaps even self-critical. Qualities we do not typically associate with large language models (LLMs). And yet, it is precisely this kind of answer that invites deeper questions: what does thinking “in the human sense” actually mean? What are consciousness, self-awareness, understanding, lived experience? And, ultimately, what is Critical Thinking?

CT Lab is a new philosophy-driven professional course at UWE Bristol, designed for professionals and researchers from any background who are willing to engage seriously with these questions and refine their critical thinking in practice. This thought piece is intended as a conceptual appetiser – an invitation to join a wider conversation.

As the opening paragraph suggests, LLMs operate through patterns. Their unmatched analytical capacity allows them to process vast quantities of data in seconds. Yet what they prioritise is not truthfulness – understood as factual, verifiable accuracy – but volume and recurrence. Patterns emerge where data is densest, most repeated, most statistically prominent. Responses are then generated accordingly.

The result is often convincing. Answers appear solid because they reflect a convergence of many sources saying similar things in slightly different ways. But plausibility is not the same as truth.
In this sense, LLMs are mimetic machines. They reproduce the surface structure of human reasoning by drawing on immense bodies of human-generated data. Consider the image of a parrot learning to speak: repetition, imitation, gradual approximation. Now imagine a parrot capable of absorbing billions of conversations within minutes. The result begins to resemble something like ChatGPT. Like that parrot, LLMs optimise for plausibility. They generate responses that align as closely as possible with dominant patterns of thought. Their aim is not to advance knowledge, but to reproduce what is already most widely expressed.

This tendency has been highlighted by researchers across disciplines. Walter Quattrociocchi, professor of computer science in Rome, and author (among many other things) of an interesting piece on “What we risk when we confuse AI and human intelligence”, described in a recent post on Facebook an experiment in which he provided an AI model with epidemiological data from 2010–2014 and asked it to construct a graph. The output included values from 2020–2021—the period of peak data density during the pandemic. The system did not “misread” the data; rather, it defaulted to the most information-rich region of its learned landscape. It followed density, not instruction.

Volume, density, reliability – LLMs are designed to play it safe. They reassure by default. And this is precisely where risk emerges. The problem is not AI itself, but our growing tendency to equate plausibility – which comforts – with truth, which often unsettles.

Critical Thinking does the opposite. It teaches us to recognise that truth is often implausible, uncomfortable, or counterintuitive. It resists the gravitational pull of consensus. Critical thinkers question what appears settled; they interrogate what feels obvious.

This requires specific intellectual dispositions. At least five stand out.

Observation: the disciplined attention to evidence, free from unexamined assumptions.
Scepticism: a readiness to doubt and to demand justification.
Curiosity: the drive to pursue knowledge beyond immediate answers.
Patience: the recognition that understanding is slow, partial, and always evolving.
Courage: the willingness to challenge accepted views, even at personal or professional cost.

These are not abstract virtues. They are practical skills – habits of mind that must be cultivated and refined.

In an age increasingly shaped by AI, Critical Thinking becomes a form of intellectual shelter – one that requires constant maintenance. CT Lab is built around this premise. The course brings together professionals from fields such as philosophy, leadership, law, policy, and strategy into a collaborative learning environment.

Across ten live sessions, participants will engage in structured dialogue with peers, guest speakers, and educators. Rather than traditional lectures, the format emphasises exchange and reflection. A curated set of multimedia materials will support the sessions, offering ongoing access to key thinkers, methods, and tools. Participants will also contribute their own real-world cases, working collectively to sharpen their critical practice.