From Invisible Algorithms to Conversational LLMs
When OpenAI announced ChatGPT on November 30, 2022, it ignited a technological revolution that pushed artificial intelligence into the cultural mainstream. In less than four years, large language models (LLMs) have become almost entirely synonymous with the term AI itself. Yet, artificial intelligence was already deeply embedded in daily life long before conversational chatbots arrived on the scene.
Before the generative AI boom, AI was mainly invisible. Algorithms performed tasks rather than engaging in open-ended chat with users. Recommendation engines on platforms like Amazon, Netflix, or social networks, translation tools, spam filters, GPS routing software, and photo classification networks all relied on specialized neural architectures. Many of these foundational models were accessible only to researchers.
The history of AI stretches back decades. Warren McCulloch and Walter Pitts designed the first artificial neural network in 1943, laying mathematical groundwork that preceded the Turing test proposed in 1950. The term “artificial intelligence” is attributed to John McCarthy, during the Dartmouth conference in 1956. Pop culture also integrated these concepts early on, famously imagining HAL 9000—also called CARL 500—in the 1968 film 2001: A Space Odyssey. However, it was not until the public deployment of ChatGPT that machine intelligence became genuinely tangible for average consumers, who could suddenly converse with a system directly.
The Dangers of Cognitive Delegation and Everyday Blunders
This accessibility has precipitated a profound behavioral shift. Users increasingly surrender everyday decision-making to generative text interfaces, a phenomenon known as cognitive delegation.
Consider a documented real-world incident involving an educated adult who decided to roast a chicken. The individual asked a chatbot for a recipe, which contained several errors. Even when a bystander with actual cooking experience pointed out the discrepancies, the user chose to follow the AI’s guidance blindly. When the chatbot subsequently suggested using a probe to verify doneness, it instructed the user to simply place the probe into the poultry cavity rather than inserting it into the meat. This improper placement yielded an inaccurate internal temperature reading.
Only human intervention prevented a culinary disaster. Yet, this anecdote illustrates a growing phenomenon: outsourcing cognitive effort makes people prone to fundamental errors, much like motorists who follow GPS systems blindly and drive their vehicles into water.
Linguistic Homogenization, Bias, and the Decline of Reading
Beyond individual decision-making errors, widespread reliance on generative models drives a broader cultural and intellectual leveling. While AI can assist in improving creativity, it reduces it at the scale of humanity. Even seemingly minor tools—such as the AI Overview recently activated in France in Google searches—fundamentally alter how people acquire information. Asking a search engine for a synonym now yields an immediate AI-generated answer, eliminating the need to browse search results.
Furthermore, LLMs encode inherent linguistic and cultural biases. Over time, heavy reliance on these models threatens to affect how humans express themselves, steering them toward specific vocabulary choices, expressions, and syntactic structures, even when they are not actively using a chatbot. Younger generations of artists growing up surrounded by synthetic media risk being influenced without realizing it.
This systemic offloading extends directly to reading habits. Reading has long served as an intimate, foundational practice for cultivating critical consciousness. The rise of LLMs capable of summarizing, analyzing, and comparing texts instantly accelerates this decline. By letting algorithms digest literature and reports on our behalf, society risks losing the essential cognitive engagement required to evaluate complex arguments.
The Long-Term Societal Reckoning
Sam Altman recently sparked intense global debate by declaring that technological singularity has begun, asserting that the boundary between artificial intelligence and human intelligence is actively dissolving.
The introduction of ChatGPT marked a profound technological and sociological turning point. It will take years, if not decades, to fully measure the consequences and determine whether the benefits outweigh the disadvantages.
What This Means for Digital Literacy
Preserving Critical Thinking: Relying on LLMs for basic calculations, culinary instructions, or document summaries diminishes personal problem-solving capabilities.
Recognizing Linguistic Uniformity: Synthetic text models introduce recurring stylistic biases that can subtly homogenize human writing and artistic expression.
Maintaining Deep Engagement: Active reading and independent analysis remain essential safeguards against automated misinformation and cognitive complacency.
