ChatGPT

ChatGPT launched on November 30, 2022, as a free research preview from OpenAI. Built on top of GPT-3.5, a model in the GPT-4 lineage trained on vast quantities of internet text, it was fine-tuned using a technique called Reinforcement Learning from Human Feedback (RLHF) — the same approach OpenAI had refined through its InstructGPT work published earlier that year. The interface was deliberately simple: a text box, a conversation thread, and a model that responded in fluent, coherent prose on almost any topic.

The technical architecture underlying ChatGPT was not new in November 2022. The transformer architecture had been introduced in 2017, GPT-3 had demonstrated large-scale language modeling in 2020, and RLHF had been applied to language models in published research since at least 2022's InstructGPT paper. What was new was the packaging: a consumer-facing product with no API key required, no technical setup, and behavior calibrated through human feedback to be helpful, harmless, and honest. The model learned to decline certain harmful requests, admit uncertainty, and maintain conversational context across a session.

The public response was unprecedented for an AI system. ChatGPT reached one million users within five days of launch and approximately 100 million monthly active users by January 2023 — a growth rate faster than any consumer application in history at that time, surpassing TikTok's trajectory. Within months, Microsoft announced a multibillion-dollar investment in OpenAI and began integrating the underlying technology into Bing search. Google declared an internal 'code red' and accelerated the release of its own large language model products. The AI industry, and the broader technology industry, reorganized around the capabilities ChatGPT had demonstrated were now commercially viable.

ChatGPT marked the moment artificial intelligence transitioned from a specialist research domain into a subject of general public concern, debate, and daily use. Before November 2022, most discussions of AI capability happened in research labs, academic conferences, and technology journalism. After it, questions about AI's ability to write, reason, deceive, replace workers, and transform education became mainstream political and cultural conversations. Regulatory bodies in the European Union, the United States Congress, and governments worldwide began drafting AI-specific legislation with a new urgency directly traceable to ChatGPT's visibility.

The longer-term significance lies in what ChatGPT demonstrated was possible with RLHF-tuned language models at scale: not just fluency, but apparent instruction-following, task completion, and conversational coherence sufficient to be useful to non-expert users. This validated a product paradigm — the general-purpose AI assistant — that every major technology company subsequently pursued. It also accelerated the commoditization of large language model capabilities, as competitors including Google, Meta, Anthropic, and Mistral released their own models within months, compressing what might have been a decade of gradual deployment into roughly two years.