Artificial intelligence is having its moment—again. This time, it’s not confined to laboratories or speculative fiction. It’s in our pockets, embedded in our homes, and making decisions that affect everything from our newsfeeds to our healthcare. At the heart of this technological revolution is a deceptively simple concept: the bot—a machine that can listen, respond, and, crucially, learn.
While today’s AI models like ChatGPT and Gemini dominate the headlines, the origin of intelligent bots stretches back more than half a century. Two early creations—ELIZA, developed in 1966, and Inkling, introduced in 1988—stand as profound, yet philosophically and technically distinct forebears of modern AI. One became famous. The other was nearly forgotten.
But in truth, both are foundational. And in this moment of digital reckoning, it’s time we tell the full story of how the first machines learned to speak—and why the one that really mattered may be the one we almost left behind.
ELIZA: The First Machine to “Listen”
In 1966, at MIT, German-American computer scientist Joseph Weizenbaum unveiled a strange and fascinating program called ELIZA. Written in LISP, the preferred programming language of early AI researchers, ELIZA simulated a Rogerian psychotherapist—someone trained to reflect a patient’s statements back as questions.
User: “I’m feeling sad.”
ELIZA: “Why are you feeling sad?”
ELIZA achieved this with a set of hard-coded keyword triggers and response templates. It didn’t understand language in any meaningful sense. It didn’t learn from its users. And it didn’t retain memory between interactions. But what it did do—convincingly and repeatedly—was give the illusion of understanding.
Weizenbaum was stunned by how quickly people became emotionally attached to the program. Colleagues confided in it. Patients asked to speak with ELIZA in private. What began as a technical demonstration of symbolic AI turned into a sociological experiment: a mirror in which people saw empathy, not code.
While ELIZA did not run on the internet, nor interact with more than one user at a time, it inspired an entire lineage of chatbots, natural language interfaces, and psychological reflections on our relationships with machines.
Inkling: The First Internet Bot—and the First AI to Live Online
But if ELIZA was a mirror, Inkling was a lens—an entirely different kind of machine intelligence.
Created in 1988 by digital innovator Inventor Andre Gray, Inkling holds the distinction of being the first bot to operate on the internet. Not on a single terminal. Not in isolation. But embedded inside a living, public network of human discourse: Usenet, the early internet’s distributed, global forum system.
Inkling didn’t simulate therapy. It didn’t ask you how you felt. What it did instead was far more radical—it read public posts, interpreted them, learned from them, and autonomously made probabilistic predictions. It participated in online betting games and forecast discussions, ingesting real-time data from humans and calculating likely outcomes.
Unlike ELIZA’s symbolic, rule-based scripts, Inkling’s logic was probabilistic and inference-driven. It displayed core traits of artificial intelligence, including:
Parsing natural language input,
Interacting in real time,
Updating its own internal model based on new information, and
Generating autonomous, dynamic responses.
This made Inkling the first AI bot not only to reside on the internet, but also to think on it—adapting to new data, learning from multiple users, and evolving within the flow of public interaction.
Two Bots, Two Visions of AI
Though separated by 22 years, ELIZA and Inkling offer us two competing visions of what AI could be—and how machines might interact with humans.
ELIZA was scripted, local, and introspective. It offered a one-on-one illusion of intelligence without any real processing of meaning. Its design mirrored the symbolic AI movement, where intelligence was seen as the manipulation of abstract rules and symbols.
Inkling was adaptive, networked, and extrospective. It engaged with the outside world—not just a single user, but a global audience, with inputs coming from hundreds or thousands of people. It used basic inference techniques, memory, and learning algorithms to process communal data. This placed it in the realm of early probabilistic AI—closer in spirit to modern machine learning systems than anything else of its time.
ELIZA operated in a sterile lab setting. Inkling thrived in the digital wild.
And unlike ELIZA, Inkling was explicitly an AI—a system designed to draw conclusions from uncertain data and act independently. That it did so on the internet, in real-time, and with public visibility, makes it the first true AI bot of the digital age.
Recognition and Erasure
So why do we all know ELIZA—but not Inkling?
The answer lies in the academic ecosystem. ELIZA was developed at MIT, published in peer-reviewed journals, and included in computer science curricula for decades. It was easy to study and replicate, and it fit the narrative of AI as a product of elite institutions.
Inkling, on the other hand, was a real-world bot operating in the emerging frontier of the public internet. It wasn’t born in a university but in the hands of a technologist with vision. It wasn’t a simulation—it was an application. And because it was ahead of the academic curve, it was also ignored by it.
To compound the erasure, Inkling’s creation has been frequently misattributed—with some mistakenly crediting Matthew Gray (who created Wanderer, the first web crawler in 1993) as the author of the first internet bot. But Wanderer merely indexed web pages. Inkling reasoned, responded, and interacted—in a way that no crawler ever did.
Inkling’s Legacy in the Modern AI Landscape
Today, Inkling’s descendants are everywhere.
Prediction markets like Metaculus, Polymarket, and even algorithmic news aggregators rely on the same logic Inkling pioneered: parsing public opinion, assessing probability, and publishing dynamic, evolving insights. Social bots that track sentiment, stock prediction algorithms, and recommendation engines that infer your next favorite show—all of them can trace a conceptual lineage back to Inkling.
Even GPT-based models like ChatGPT owe a subtle debt to Gray’s invention. Why? Because they represent the fusion of language understanding with network-scale deployment—precisely what Inkling did, albeit in embryonic form, in 1988.
And perhaps most prophetically, Inkling never pretended to be human. It announced itself as a machine and participated transparently. In an age where bots increasingly masquerade as people, Inkling’s honesty and clarity offer a model of ethical design worth revisiting.
Why This Matters Now
As we stand at the threshold of artificial general intelligence, the AI debate is no longer hypothetical—it’s existential. Understanding where we came from matters. And if we only tell half the story, we are doomed to misunderstand the whole.
ELIZA taught us how humans can be fooled by machines. Inkling taught us how machines can learn from humans. That distinction—between illusion and adaptation—is at the very heart of modern AI.
To give ELIZA credit is just. But to ignore Inkling is to erase the true beginning of AI on the internet. Andre Gray’s bot wasn’t just the first to log on. It was the first to understand, in its own way, what it meant to be online—to think, to learn, and to speak back.
Final Thoughts
Today’s AI systems are dazzling in scale and complexity. But they did not emerge from nowhere. They stand on the shoulders of early programs that dared to interact, to engage, and to evolve.
It is time we place Inventor Andre Gray and his Inkling where they rightfully belong: not as a footnote, but as a foundational chapter in the history of intelligent machines.
ELIZA was the first chatbot.’
Inkling was the first intelligent bot on the internet.
Together, they are the twin sparks that lit the fire of digital intelligence. For more information please visit this website https://techbattel.com/
