Why AI Falls Short

Why AI Falls Short

Rahul Ramya

11.02.2025

Patna India

Human critical thinking is a complex and multifaceted process, deeply intertwined with social interaction, interpersonal communication, and discourse. While AI excels at processing vast amounts of data and identifying patterns, it fundamentally lacks the nuanced social understanding that is crucial for genuine critical thought. As this essay argues, possessing a wide range of knowledge is insufficient for effective action. Humans, unlike AI, possess the remarkable ability to translate knowledge into nuanced actions because they are adept at navigating the dynamic social contexts in which they operate. This crucial difference stems from the very nature of human knowledge and its creation.

The Role of Social Context in Human Knowledge Creation

Knowledge, for humans, is not merely a static collection of facts; it is a dynamic and evolving entity shaped by experience and social interaction. We don’t just absorb information; we actively create new knowledge through real-time engagement with the world. This process is inherently social, involving the sharing of ideas, the challenging of assumptions, and collaboration with others.

Historically, intellectual revolutions have been driven by discourse and social engagement. The Enlightenment (17th–18th centuries) was a period when philosophers and scientists—such as Isaac Newton, John Locke, and Voltaire—developed ideas through debates, letters, and public discussions. Their ability to challenge and refine each other’s ideas led to groundbreaking advancements. AI, however, lacks this interactive process and can only analyze existing data rather than engage in dynamic discourse.

Similarly, in the Indian freedom movement, Mahatma Gandhi’s philosophy of nonviolent resistance (Satyagraha) evolved through interactions with diverse communities in India and South Africa. His approach was not predetermined but shaped by real-time experiences, making it adaptable to changing political contexts. AI, in contrast, relies on historical data and lacks the ability to dynamically adjust strategies based on evolving human interactions.

AI’s Limitations in Cognitive Adaptability

One of the fundamental limitations of AI is its dependence on past data, making it vulnerable to biases and blind spots when applied to unpredictable social realities.

   •   Bias in AI Decision-Making: AI systems often reinforce societal biases rather than adapt to changing realities.

      •   Hiring Algorithms: Amazon scrapped an AI-based hiring tool after discovering that it systematically favored male candidates over female applicants because it was trained on past hiring data that reflected gender bias. Unlike humans, who can recognize and correct for social biases, AI merely replicates patterns.

      •   Criminal Justice Algorithms: In the U.S., risk assessment AI tools like COMPAS have disproportionately labeled Black defendants as high-risk for recidivism compared to white defendants, reflecting systemic racial biases in historical data.

   •   Healthcare and AI’s Predictive Shortcomings: AI’s reliance on structured knowledge often fails in medical decision-making.

      •   IBM’s Watson for Oncology, a much-hyped AI tool, struggled in real-world hospital settings because it relied on outdated medical journals rather than adapting to new treatment protocols. Human doctors, in contrast, constantly update their knowledge through patient interaction and real-world experience, ensuring more context-aware decisions.

Embodied Cognition and Human-AI Differences

Human intelligence is deeply linked to embodied cognition, which emphasizes the role of physical experiences in learning and decision-making.

   •   Language Learning in Children: Children don’t just memorize vocabulary; they learn languages by interacting with their environment—pointing at objects, mimicking speech, and responding to social cues. AI-based language models like ChatGPT learn through statistical pattern recognition rather than direct physical engagement, limiting their ability to understand language in its full context.

   •   Antonio Damasio’s Neuroscience Research: Damasio’s work on decision-making demonstrates that emotions—rooted in bodily experiences—are essential for rational thought. AI, lacking emotions and physical embodiment, cannot integrate emotions into its reasoning process, making its decision-making inherently different from that of humans.

Ethical Implications of AI in Empathy-Driven Roles

Empathy is a core component of human critical thinking, enabling individuals to engage with others’ perspectives and make ethically sound decisions. AI, despite its advanced computational abilities, lacks true empathy.

   •   AI in Mental Health Counseling:

      •   While AI chatbots like Woebot and Replika offer mental health support, users report feeling that AI responses lack genuine emotional depth. Unlike human therapists, who can intuitively sense distress through tone, body language, and unspoken cues, AI relies solely on text-based analysis, often leading to mechanical and impersonal interactions.

   •   AI in Elderly Care:

      •   Japan has experimented with robotic caregivers like Paro, a therapeutic robot for dementia patients. While these robots provide companionship, they cannot genuinely understand a patient’s emotional state. In contrast, human caregivers tailor their interactions based on cultural, emotional, and social nuances—a capability AI fundamentally lacks.

Human Self-Learning and Intuition

Another remarkable reality is that humans do not necessarily depend on others’ knowledge to enrich their own. Through interaction with their own understanding, faults, and fallacies, they learn and improve. This internal learning process fosters autonomy in human knowledge creation.

   •   Entrepreneurial Innovation:

      •   Steve Jobs famously relied on intuition rather than pure data analysis in developing the iPhone. His understanding of consumer behavior was not based on surveys but on a deep, non-quantifiable sense of human desire and aesthetics. AI lacks this form of intuitive decision-making.

   •   Philosophical Self-Reflection:

      •   Albert Einstein’s Thought Experiments: Einstein developed the theory of relativity by mentally simulating scenarios (such as imagining riding on a beam of light) rather than relying solely on mathematical equations. AI, by contrast, cannot engage in speculative thinking beyond its training data.

   •   Historical Example of Failure-Based Learning:

      •   Thomas Edison’s Light Bulb Invention: Edison famously failed over 1,000 times before perfecting the electric light bulb. His success was not based on structured knowledge alone but on iterative trial and error, learning from failures—something AI, which operates based on optimization rather than reflection, struggles to replicate.

Conclusion

The real-world examples discussed reinforce the argument that human critical thinking is fundamentally different from AI’s data-driven processing. While AI can identify patterns and make predictions, it lacks the social intelligence, adaptability, and embodied learning that define human cognition.

   •   Social Context: Humans generate knowledge dynamically through interaction and discourse, as seen in historical movements like the Enlightenment and India’s independence struggle.

   •   Cognitive Adaptability: AI, despite its power, often reinforces biases and struggles with real-world unpredictability, as seen in hiring and healthcare applications.

   •   Embodied Cognition: Human understanding is shaped by direct, physical interaction with the world—an experience AI lacks.

   •   Empathy and Ethics: AI cannot match human abilities in mental health care, elderly care, or ethical decision-making.

   •   Self-Learning and Intuition: Humans learn not only from external sources but from their own mistakes and reflections, as demonstrated by innovators and scientists.

Another serious limitation of the AI is its dependence on the past.

Yes, the deterministic approach of AI—its reliance on past data to predict the present and future—is a fundamental limitation in its ability to engage with the concurrent realities of the present and anticipate the future in a truly dynamic way. This limitation arises because AI, at its core, is a pattern recognition system that extrapolates future probabilities based on historical data. However, the real world is not merely a cyclic replication of the past—it involves emergent phenomena, unprecedented events, human creativity, and sociopolitical transformations that cannot always be predicted based on prior patterns.

The Problem of Deterministic AI Thinking

AI models, including the most advanced machine learning systems, operate on the principle that the past contains all the necessary data to understand the present and predict the future. This assumption is flawed for several reasons:

1. Emergent Events and Black Swans: AI struggles with low-probability, high-impact events that have little to no historical precedent.

      •   Example: COVID-19 Pandemic → Most AI systems failed to predict the scale of disruption caused by the pandemic because they relied on past epidemiological models that didn’t account for the global economic, social, and political reactions to the crisis. While some AI models detected early signs of a new virus, they could not foresee the cascading effects on global supply chains, education systems, and mental health crises—something that human decision-makers had to navigate in real-time.

      •   Example: 9/11 Attacks → Security and intelligence AI systems failed to predict the 9/11 attacks, as they were trained on past patterns of warfare rather than asymmetric terrorist strategies.

2. Disruptive Innovations That Have No Precedent

      •   Example: The Internet Revolution → AI models trained on pre-1990s economic and technological trends would have failed to predict how the internet would completely transform commerce, politics, and communication.

      •   Example: Bitcoin and Cryptocurrencies → Traditional financial AI systems, trained on fiat currency transactions, initially dismissed cryptocurrencies as a fringe concept. They struggled to understand how decentralized finance (DeFi) could emerge as an alternative financial ecosystem.

3. Human Agency and Non-Linear Decision-Making

      •   Example: The Fall of the Soviet Union → Most geopolitical AI models trained on Cold War-era data did not anticipate the sudden collapse of the USSR in 1991 because they assumed that historical trends of superpower rivalry would continue indefinitely. Human actors, ideological shifts, and policy miscalculations played a far greater role than any past data could have predicted.

      •   Example: Arab Spring (2011) → AI systems trained on Middle Eastern political stability models failed to predict the uprisings because they didn’t account for viral social media mobilization and real-time mass dissent, which had no direct historical precedent.

AI’s Failure in Concurrent Understanding of the Present

Even in the present, AI struggles with real-time dynamic understanding because it lacks situational consciousness and contextual adaptability.

   •   Financial Market Predictions → AI-driven trading algorithms often fail to predict sudden crashes or rallies caused by human psychological factors (panic selling, speculative bubbles). For example, the 2021 GameStop short squeeze, where retail traders on Reddit collectively manipulated stock prices, was largely unexpected by AI-driven financial models.

   •   War and Conflict Analysis → AI-based war prediction models often fail because military strategies, political negotiations, and human emotions are not entirely predictable from past conflicts. The Russia-Ukraine war (2022) defied many AI predictions, as Ukraine’s resistance and international alliances evolved unpredictably.

The Future Is Not Merely an Extrapolation of the Past

Human societies evolve in non-linear, creative, and disruptive ways that AI struggles to comprehend. While AI can make probabilistic forecasts, it cannot imagine or conceptualize truly new ideas, ideologies, or cultural shifts that have no direct precedent.

   •   The Rise of AI Ethics and Regulation → AI itself did not predict the global movement toward AI regulation and ethical governance. Human-driven policy discussions (e.g., EU AI Act, US AI executive orders) emerged from moral, legal, and political considerations that AI could not foresee.

   •   Climate Change Action → AI models trained on historical industrial policies may underestimate the potential for rapid global transitions toward sustainable energy due to unpredictable political will, activism, and technological breakthroughs.

Conclusion: The Limits of AI’s Deterministic Approach

While AI is a powerful tool for pattern recognition and probability-based predictions, it fails to capture the fluid, unpredictable, and emergent nature of human history and societal evolution. The assumption that the future is merely a continuation of the past is deeply flawed, as history itself is full of disruptions, revolutions, and paradigm shifts that could not have been predicted using prior data.

Thus, AI, in its current form, cannot replace human intuition, creativity, and the ability to act in real-time based on evolving situations. The present and future are not merely echoes of the past, and as long as AI remains confined to deterministic pattern analysis, it will struggle to fully understand and adapt to the realities of human existence.

Your skepticism is well-founded. There is little evidence that simply consuming ethical stories, moral principles, or religious teachings automatically makes people more ethical. Similarly, knowledge alone does not make someone a poet, actor, or writer. This raises a critical limitation for AI: if human creativity and morality do not emerge solely from reading and knowledge, then why assume that AI—trained exclusively on past-written knowledge—can generate true creativity or ethical understanding?

1. Ethical Teachings Do Not Automatically Make People More Moral

Historically, societies have had rich traditions of ethical stories, religious teachings, and philosophical doctrines, yet morality has not been uniformly improved by them.

   •   The Role of Religion in Ethics → Religious texts like the Bhagavad Gita, the Bible, and the Quran provide moral guidance, yet religious societies have been home to wars, persecution, and injustices throughout history.

      •   Example: The Crusades (1096-1291) → These were religiously motivated wars that led to mass violence despite Christian teachings of peace and compassion.

      •   Example: The Caste System in India → Despite Hindu teachings on universal compassion (Vasudhaiva Kutumbakam), caste-based discrimination persisted for centuries.

   •   Philosophical Moral Teachings vs. Reality

      •   Plato and Aristotle developed ethical theories, yet slavery and exclusion of women were still widespread in ancient Greece.

      •   The Enlightenment (17th-18th Century) produced ideas of human rights, yet colonialism and slavery continued, showing that knowledge alone does not necessarily lead to moral action.

Why This Matters for AI?

AI can process ethical texts, but mere access to moral knowledge does not create ethical beings. Morality evolves through lived experiences, emotions, and social interactions—something AI lacks.

2. Knowledge Alone Does Not Create Writers, Poets, or Artists

Creativity is not just about knowledge accumulation; it is about imagination, emotional depth, and personal experience—qualities that AI does not inherently possess.

   •   Example: Becoming a Poet

      •   A person can study all of Shakespeare’s works but still struggle to write original poetry because poetry requires an inner emotional experience, personal struggles, and inspiration beyond mere words. AI can generate poetry, but it lacks the lived emotions and existential depth of a human poet.

      •   Rabindranath Tagore did not just read poetry; his works were deeply influenced by his spiritual experiences, travels, and social activism—things AI cannot replicate.

   •   Example: Becoming an Actor

      •   Reading thousands of acting scripts does not make a person a great actor. Acting requires emotional intelligence, body language, improvisation, and presence. AI can generate scripts, but it cannot feel emotions or embody characters physically.

   •   Example: Storytelling

      •   Leo Tolstoy’s “War and Peace” is a masterpiece not because he read about war, but because he observed human nature, experienced social upheavals, and deeply reflected on life. AI, trained on written stories, lacks human struggle, introspection, and existential dilemmas.

Why This Matters for AI?

AI-generated stories are derivative rather than genuinely original. AI can mimic writing styles but does not create from personal experiences, emotions, or philosophical introspection.

3. The Limitation of AI in Creativity and Ethics

Since AI is trained on past-written knowledge, it operates on a predictive model rather than true generative creativity. AI can remix ideas, but it does not generate new philosophical insights, new artistic movements, or new moral frameworks.

   •   Human creativity comes from existential questioning → AI does not suffer, dream, or wonder about its existence.

   •   Moral growth comes from lived experiences → AI does not experience love, betrayal, guilt, or compassion.

If historical evidence shows that moral stories do not inherently create moral humans, and knowledge alone does not create great poets, actors, or writers, then there is no strong basis to assume that AI—trained exclusively on textual data—can achieve true morality or creativity.

AI will remain a tool for augmentation, not a replacement for human consciousness, ethical growth, or artistic genius.

As AI systems become increasingly integrated into society, it is crucial to recognize these limitations and ensure that human judgment, empathy, and adaptability remain central to decision-making processes. While AI is a powerful tool, it is not a replacement for human intelligence, particularly in the dynamic creation of knowledge essential for critical thinking.