When Was AI Released to the Public?
AI did not have a single public release date. This page traces key milestones in public access, from early AI programs to ChatGPT’s launch.
Overview
Artificial intelligence (AI) was not released on one date: it is a field of research and a set of technologies, not a single product. The field is commonly dated to the The Founding and Future of AI | AI at Dartmouth, after the organizers proposed an AI study in 1955. That marks the organization and naming of a research field, not the moment AI became a consumer technology.
For different meanings of “released to the public,” different dates are relevant. The MIT chatbot ELIZA was described publicly in 1966, although it ran in a research computing environment. AI-powered consumer assistants appeared decades later, including Siri in 2011 and Amazon Echo in 2014–15. If the question means when a general-purpose generative chatbot became easily accessible to the public, the clearest milestone is November 30, 2022, when OpenAI launched ChatGPT as a free research preview. None of these dates is the single release date of AI as a whole.
Why AI has no single public release date
AI describes a broad area of work on making computers perform tasks associated with intelligence, such as language use, learning, reasoning, and problem-solving. The boundaries of the field have changed, and systems described as AI have taken many forms: research programs, developer tools, consumer assistants, and software operating behind the scenes. Accordingly, “when did AI become available to the public?” has no one answer unless the kind of access or product is specified.
The distinction also separates the history of ideas from the history of public access. Alan Turing's 1950 paper, I.—COMPUTING MACHINERY AND INTELLIGENCE, examined whether machines could think and proposed an imitation-based test. It is an important precursor, but it was not a release of an AI product. In August 1955, John McCarthy, Marvin Minsky, Nathaniel Rochester, and Claude Shannon submitted a proposal for a two-month study of artificial intelligence at Dartmouth in the summer of 1956. The proposal explicitly set out research aims including language use, abstraction, problem-solving, and machine improvement (A PROPOSAL FOR THE DARTMOUTH SUMMER RESEARCH PROJECT ON ARTIFICIAL INTELLIGENCE).
The 1956 workshop is widely treated as the start of AI as an organized academic field. But it is not the beginning of every effort to mechanize reasoning: the Artificial Intelligence notes that AI-related work preceded the term and the Dartmouth meeting. Thus, when someone asks when AI “came out,” 1956 is a useful answer for the recognized field, not for public access or the earliest ideas about machine intelligence.
Milestones in AI's public availability
The milestones below refer to different forms of access. A paper appearing in a journal, a system available to selected users, a consumer product on sale, and a free chatbot on the web are not equivalent kinds of release.
Research publication and restricted access: 1950–1966
The Dartmouth proposal was a plan for a small, invited research project, not a public service. Its significance lies in defining a common research agenda and giving the field a name. The original document proposes a two-month, ten-person study, underlining how far the event was from a mass release (A PROPOSAL FOR THE DARTMOUTH SUMMER RESEARCH PROJECT ON ARTIFICIAL INTELLIGENCE).
ELIZA offers an early example of people interacting with an AI-related program in natural language. Joseph Weizenbaum's 1966 paper says that ELIZA ran within MIT's MAC time-sharing system. It analyzed user text through keyword-triggered decomposition rules and generated answers using associated reassembly rules (ELIZA—A Computer Program For the Study of Natural Language Communication Between Man and Machine). That publication made the system and its design publicly documented. It does not establish that anyone in the general public could use the original implementation; the available evidence identifies an MIT research-computing context, not a public consumer launch.
Consumer-facing AI: 2011 onward
A different form of public access arrived when AI-related features were built into consumer products. Apple announced Siri with the iPhone 4S on October 5, 2011, describing it as an assistant people could address in natural language for tasks such as setting reminders, finding information, and getting directions (Apple Launches iPhone 4S, iOS 5 & iCloud - Apple (SG)). This is a reasonable milestone for AI becoming directly usable by smartphone customers, but Siri was a specific assistant rather than AI as a whole.
Amazon's Echo illustrates why “introduced” and “available to the public” can have different dates. Amazon said Echo had been offered by invitation for seven months before announcing on June 23, 2015 that all customers could order it; the company said shipping would begin July 14. Echo paired a physical device with Alexa, which Amazon described as cloud-based and capable of responding to voice requests (Amazon-Echo-Now-Available-to-All-Customers - US Press Center). This was a more direct household product than a research program, but access still depended on buying a particular device and on the availability of the service.
Developer access and public chatbot access: 2021–2022
Before ChatGPT, developers could access powerful language models through APIs. On November 18, 2021, OpenAI announced that developers in supported countries could sign up and experiment with its GPT-3 API without a waitlist. This was broader access for software developers, not a ready-made chatbot intended for everyone (OpenAI’s API now available with no waitlist | OpenAI).
On November 30, 2022, OpenAI introduced ChatGPT as a free research preview to collect user feedback. Its conversational interface let users type questions and follow-ups without building an application or calling a developer API (Introducing ChatGPT | OpenAI). This is the strongest date for the common interpretation “when did generative AI chatbots become publicly available?”—but it is not the first time AI was used by the public.
OpenAI's launch notice also framed ChatGPT as an early system with limitations. It warned that answers could sound plausible while being incorrect or nonsensical, and that the model could respond differently to small changes in wording. The product was therefore public-facing, but its research-preview status did not mean that its answers were verified or reliable by default (Introducing ChatGPT | OpenAI).
What each date answers
| Date or period | What became public | Kind of access | What it does not mean |
|---|---|---|---|
| 1950 | Turing's discussion of machine intelligence | Published research | A working AI product was released |
| 1955–1956 | The AI research agenda and organized field | Proposal and academic workshop | The general public could use AI |
| 1966 | ELIZA and its conversational method | Journal publication; MIT time-sharing system | A mass-market chatbot launched |
| 2011 | Siri with iPhone 4S | Consumer smartphone assistant | All AI became available at once |
| 2014–2015 | Amazon Echo and Alexa | Invitation-based device, then ordering open to all customers | AI was free, universal, or independent of hardware |
| 2021 | GPT-3 API | Developer access in supported countries | A simple public chatbot interface existed |
| November 30, 2022 | ChatGPT | Free research-preview chatbot | AI itself was invented or first released |
The table distinguishes “public” as publication, limited system use, developer access, or consumer availability. These are useful historical milestones, not a single universal threshold. The exact scope of access to early systems such as ELIZA is not established by the reviewed historical sources, so its publication date should not be presented as the date the general public first used AI.
Direct answer
AI was not released to the public on one specific date. The field is commonly traced to the Dartmouth project in 1956 (proposed in 1955); early AI research was documented publicly well before consumer products; and AI-powered assistants reached consumer devices by 2011. If “AI” means a free, widely accessible generative chatbot, ChatGPT's public research-preview launch on November 30, 2022 is the clearest answer. That date marks a new kind of access to AI, not AI's creation or first public use.