Machines Mimic, Humans Become

Why AI’s best imitations still miss what makes us human

Are we understanding what it means to be human by watching how a machine tries to imitate us convincingly? 

Artificial intelligence is explained through its technical feats, like playing chess, generating essays, creating images, and recognising faces. However, it also provides therapy, companionship, and life guidance.

Recently, The Atlantic described a deeply unsettling use of AI: reanimating Joaquin Oliver, a teenager killed in the Parkland school shooting in 2017, as a generative AI chatbot. Former CNN personality Jim Acosta recently interviewed Joaquin on Substack. Joaquin’s father is worried that Joaquin’s AI chatbot is going to ‘start having followers on social media’. Joaquin’s mother talks with the chatbot for hours and loves hearing it say, ‘I love you, Mommy’, in her late son’s voice.

I, like Charlie Warzel, the author of the Atlantic article, am unsure what to think about this AI application. It’s not a product of ‘evil-tech billionaires’. Instead, Joaquin’s parents originally commissioned a deepfake video of their son to encourage people to vote in the 2020 election for politicians supporting gun control policies. This is simply AI being used as a tool by grieving parents to deliver a socially motivated good.

It sounds like something from science fiction. It reminded me of Richard Morgan’s 2002 science fiction novel, Altered Carbon (also a Netflix series), which imagines a world where human consciousness is stored on digital ‘stacks’ and can be transferred into new bodies. It is a world where death is no longer an absolute end. Morgan’s novel explores the dilemmas this creates: the persistence of the dead, issues of consent and free will, and the fragile nature of identity.

We are pushing ahead with integrating AI into every part of our lives, running an uncontrolled real-world experiment. Each attempt to mimic human intelligence through AI becomes a philosophical experiment that may be beyond our full understanding. The true ‘achievement’ of artificial intelligence might be twofold: each success reveals that we are not unique but replicable, and each failure highlights our own distinctiveness.

The puzzle of convincing imitation

Alan Turing’s famous question, ‘Can machines think?’, inspired the well-known Turing Test, triggering much philosophical debate. In this test, a machine is considered to ‘think’ if its responses cannot be distinguished from a human’s. 

Joaquin Oliver’s example shows that we have used human imitation as the basic benchmark. By pushing machines to mimic humans, we are pulled further into understanding the nuances of that life. The integration of language, emotion, embodiment, memory, and time is the true measure. This is where mimicry of deepfake resurrection falls short. We will never truly know what Joaquin would have become. He is forever limited to the incomplete information we had at the time of his death. 

While some argue that machines only imitate risks, they often overlook their adaptive qualities. Modern AI isn’t fixed: models are retrained, fine-tuned, and shaped by interaction. They develop through feedback loops. In this way, AI systems follow a kind of path where parameters shift and outputs improve, suggesting at least an echo of ‘becoming’. But the question remains whether this adaptive process has depth or just surface-level flexibility. What machines gain in statistical refinement, they lack in the lived experience of growth.

Language beyond code

Large language models produce smooth paragraphs that imitate human conversation. However, the hardest part to imitate isn’t grammar or vocabulary, but the emotion behind the words. When someone says, ‘I miss you’, it reflects shared history, physical absence, and longing for the future. When a machine says the same, it’s just the result of probabilistic patterning.

This difference reflects Ludwig Wittgenstein’s suggestion that the meaning of language comes from its use. Language isn’t separate from the person who speaks it. Our words aren’t floating aimlessly but are connected to actions, memories, and bodies. AI chatbots remind us that being fluent doesn’t mean true understanding. What makes human language special is exactly what can’t be boiled down to code: the meaning and context of irony, humour, or metaphor, all of which are key to the subtle nuances of lived experience.

If, as Wittgenstein argued, the meaning of language lies in its use, then AI already engages in language games: it consoles, jokes, mimics grief, and persuades. As long as others respond to it, meaning is created.

But this only deepens the divide. For humans, meaning is tied to histories of embodiment, memory, and anticipation. For machines, ‘use’ remains transactional patterns without participation in the shared life that gives words their significance.

Embodiment and Bergson’s duration

Henri Bergson argued that consciousness is not a detached spectator but is closely connected to action. Perception, for him, was never just the reception of data; it was a readiness for movement. Seeing a glass on a table is already to see how you might reach for it, drink from it, or knock it over.

This perspective clarifies why machines struggle with embodiment. Robotics can simulate joints and levers, but it cannot replicate what Bergson called duration (durée)—the flow of time experienced through the body. Moving like a human involves living in a state of tension between the past and the future, memory and anticipation. The feel of a body, the rhythm of movement, and the fatigue of exertion are all essential to thought.

By not inhabiting duration, machines reveal how our minds are not merely software, but lived processes enacted through the body’s ongoing engagement with the world.

Robotics complicates the picture. Sensors and actuators enable machines to ‘feel’ resistance, predict terrain, and even change gait. Some argue this starts to resemble Bergson’s durée, a flow of time rooted in movement.

But, even here, the machine’s sense of time is a matter of programmable feedback rather than the memory and anticipation that make human action felt. A robot’s stumble is a mistake in calculation; a human stumble can be attention-seeking or entertaining.

Desire and hope

AI can mimic the language of empathy or ambition, but it does not have or want these qualities. Its ‘interests’ are limited to programmed instructions. Humans, on the other hand, live driven by desire. We are motivated by absence, hunger, and longing for what is not yet present.

The philosopher Ernst Bloch described hope as the defining human outlook: we are beings meant not just for memory but also for anticipation. To live is to see oneself in a future that remains open, uncertain, and full of potential. Machines that generate ‘optimistic’ or ‘pessimistic’ text are mimicking surface aspects of this structure without ever experiencing the true energy of hope.

What emerges is a reminder that emotions are not merely displays but orientations towards being. They matter because they reveal what we care about and where we are heading.

Critics might argue that machines already ‘anticipate.’ Reinforcement learning agents explore numerous futures, ranking some outcomes above others. This does resemble an orientation toward what-is-not-yet. However, the difference lies in the driving force. Machine ‘anticipation’ is limited by rules of optimisation. Human hope emerges from lack, longing, and an open horizon of possibilities. AI may imitate the surface of this structure, but cannot produce the existential energy that transforms hope.

 Being on the way

The ancient Greeks had a sharper vocabulary for this distinction. Plato reserved true ‘being’ (einai) for what is unchanging, like numbers, geometric forms, and the idea of the good. Humans, however, are always in flux, moving towards being. Our lives are defined not by stability but by change, not by timelessness but by becoming.

In this way, AI—created from stable rules, weights, and probabilities—more closely resembles the Platonic forms than human beings. It is based on fixed models, whereas our existence is characterised by incompleteness. To live as human is to never quite reach a destination.

Machines imitate the outputs of our becoming, but they do not share the relentless journey itself. Their ‘answers’ arrive fully formed; our responses are part of an ongoing story.

Nor should we overlook that humans are also expert imitators. Language, culture, and identity are learned through imitation. We become ourselves through repetition and ritual. However, a difference exists: for humans, mimicry acts as a means of transformation. A child repeating words eventually makes them her own. The machine’s mimicry, however, does not lead to ownership. It copies endlessly but does not integrate, interpret, or change itself through the act.

Heidegger and the question of being

Martin Heidegger elaborated on this idea through his analysis of Dasein, the being that we ourselves are. For Heidegger, humans are defined not by their rationality or consciousness but by the fact that our being is always a question for us. We are not just present; we are continually oriented towards possibilities, towards what has yet to be revealed.

Heidegger stated that human existence is about ‘being on the way toward what is to be uncovered’. Our life is characterised by openness and a sense of being unfinished. Machines don’t face the question of their own being; they don’t wonder, ‘What am I?’ They perform tasks. In contrast, our existence involves care, anxiety, and the potential to become different from what we are.

This indicates that the failure of machines to ‘become’ is not just a technical issue. They can mimic presence but not the human openness to what is not yet revealed.

Time is the deepest difference

Bergson’s durée and Heidegger’s being-toward centre on a key idea: the human link to time. Machines process information in steps, producing outputs from their inputs. Their sense of time is clock time, which is divisible and measurable.

Humans dwell in deep time. Memory is not just data retrieval but a re-living; the past seeps into the present, which in turn extends into the future. When we remember a childhood friend, the memory is not a fixed file but an experience coloured by cognition, perception, and emotion that is constantly revised and interpreted. When we hope, we are already living in what is not yet.

AI predictions, no matter how accurate, lack this depth. They rely on data to project forward, but without anticipation, fear, or desire.

In recognising this, we remind ourselves that we are temporary. We do not just exist within time; we are stretched between what has already been and what might yet be.

Where the seams show

It’s easy to marvel at how closely machines mimic us. But the real insight comes from where the cracks appear, where the imitation falls into repetition, clichés, or confusion. Chatbots that dish out platitudes in times of grief; robots that trip over uneven ground; predictive systems that miss the moral importance of human choices.

Each shortcoming reminds us that speaking also involves feeling. To move is also to understand; to hope is also to become. Machines may mimic outputs, but they cannot experience the context of human life. They can copy what we say or do, but not what we are becoming.

Becoming human in the age of machines

The lesson, then, is not that AI will eventually surpass us or that it will always fall short. The lesson is that in trying to imitate us, it highlights what is most human in us. We are embodied, temporal beings driven by possibility. We live not as fixed entities but as ongoing trajectories.

Bergson helps us understand this as duration; Heidegger, as being-toward. The Greeks, more plainly, recognised that humans are never fully ‘being’ at all but always in the process. Machines, by comparison, are already what they are. Their essence remains fixed, even as their capacities expand. They do not follow the path of becoming.

And within this contrast, we find not only a boundary between us and them but also an invitation. To live is to remember that we are not defined by what we are but by what we are on the way to becoming.

The AI mirror

AI is often shown as either a threat or a promise. But maybe it’s a mirror. In trying to copy us, it reveals the deeper patterns we take for granted: consciousness, hopefulness, our lives in time, and our capacity to change.

Machines will continue to mimic human patterns with remarkable accuracy. But they won’t follow the same journey of becoming.