A Brief History Of AI technology
Is AI Technology Alive?
In June 2022, Google AI technology engineer/priest Blake Lemoine publicly declared that he believes the Google AI chatbot generator LaMDA (Language Model for Dialogue Applications), which he has been working with, has achieved sentience. If true, this is a monumental statement, though Google has been quick to reassure that its AI technology is not yet at that level. Regardless, the eye-catching headlines have sparked new discourse in the field and beyond with computer scientists and engineers recognizing that it may be only a matter of time before we cross that threshold as a society.
AI In The IKIN Blog
In the previous IKIN Blog, we looked at the foundational definitions of AI technology and specifically at four popularly identified kinds of AI technology–current and theoretical. For most in the field, Self-Aware AI remains theoretical, as does the pre-cursory level Theory of Mind AI. Thus, current AI technology remains up to and within the level of Limited Memory AI where machine learning is possible but AI remains squarely insentient. In this IKIN Blog, we’ll be tracing the introduction and evolution of AI technology to see how it has come to be that today some are already asking if AI is self-aware.
A History of AI Technology Turning Points
Laying The Groundwork For AI
Many trace the origins of AI technology to the first half of the 20th century when fantastical science fiction works, like The Wizard of Oz children’s book and film and the German Expressionist picture Metropolis, laid the conceptual groundwork for AI through robot characters. By mid-century, the famed British mathematical and inventor of modern computing Alan Turing floated the more scientifically-grounded question “Can machines think?” in his 1950 paper “Computing Machinery and Intelligence.” He followed this work by outlining the Turing Test, which is still used to this day to assess the believability of modern AI to masquerade as human. The term “artificial intelligence” was coined in 1956 at a conference at Dartmouth University, and the associated term “machine learning” was introduced a few years later in 1959 in reference to a checkers-playing machine. As these essential concepts, questions, and terms were being laid, science worked toward actualizing and improving AI.
Putting AI Into Practice
According to Britannica, the first successful AI program was designed to play checkers. It was written in 1951 by early British computer scientist Christopher Strachey. Many of the first successful AI programs were based on checkers and eventually chess, games with clearly defined rules and strategies that allowed Reactive AI programs to accumulate and apply outcomes to a multitude of defined scenarios. Stateside, IBM led the charge in developing AI technology. A year after Strachey introduced his checkers AI, Arthur Samuel and IBM introduced theirs, which was able to not only store and apply successful strategies but could learn from experience and even evolve its own programming. This learning introduced the concept of evolutionary computing, which is the foundation of the machine learning branch of AI technology. These AI game programs became so efficient at their tasks that over the decades they have surpassed even the greatest human players.
AI Is Crowned The Gaming King
Checkers
The 1990s marked a turning point for AI challenging the supremacy of the human mind in gaming. The first of these milestones came in 1994 with the AI checkers program Chinook, developed by computer scientists from the University of Alberta. Not only did Chinook beat the raining human chess champion but it perfected the game to such a level as to make it unbeatable by playing out and storing every possible move, roughly 500 billion billion in total.
Chess
On February 10, 1996, IBM’s AI chess-playing supercomputer Deep Blue finally beat Russian chess grandmaster Garry Kasparov in a game. Kasparov took the match, but a year later Deep Blue returned to claim a decisive victory in the rematch.
Go
It took another two decades for AI technology to claim the immensely complex game Go, which is calculated to have more moves than there are estimated atoms in the universe. In March 2016, Google’s DeepMind AI, running the program AlphaGo, won three decisive matches against world-renowned Korean player Lee Sedol. Even more shocking, this came nearly a decade before popular opinion on AI technology believed it would be able to achieve such a feat.
Advancing AI: Robots And Smart Programs
AI Robots
In the decades following the conceptual establishment and introduction of functional AI, its technology developed immensely and in many ways. In addition to the important game-playing AI milestones identified above, the evolution of AI technology also includes the development of early robots starting in 1961 with the Unimate, the first industrial robot. In 1968, SRI International debuted Shakey, the first general-purpose robot. And 1998 saw the introduction of Kismet, the first robot programmed to recognize and simulate human emotions.
Chatbots And Other Smart Programs
A few years before Shakey, the first chatbot was invented in 1964. Eliza, as it was known, was a psychotherapeutic robot designed to give preloaded responses to users as a kind of robo-psychiatrist. The product of Joseph Weizenbaum of the Artificial Intelligence Laboratory at MIT, Eliza set the groundwork for a long lineage of conversational AI, including Apple’s Siri, which debuted in 2011, and Amazon’s Alexa, which hit markets in 2014. These modern forms of AI technology join the many AI phone operator systems used today to direct caller inquiries to appropriate answers or to the human best suited to address an inquiry.
Watson And Deep Learning
IBM’s question-answering supercomputer Watson can also be included in the list of Eliza’s descendants. Watson, most known for its popular routing of Jeopardy’s top champions, however, also owes its genius to the development of neural networks, which were first proposed in 1944 and have ebbed and flowed in popularity since. Today, neural networks are the foundation for some of our most impressive modern AI technology thanks to the parallel development of deep learning within these systems. According to IBM, deep learning “attempts to mimic the human brain…enabling systems to cluster data and make predictions with incredible accuracy.” It’s this combination of deep learning AI, realized through the ever-increasing processing and memory storage capacities of modern computing, that have brought us to the current state of AI technology.
The Current State Of AI Technology
So here we are in a world surrounded by AI technology. Watson has moved on from its Jeopardy triumphs to offering its deep learning tech for AI business solutions. AI pilots our self-driving cars, it filters our email inboxes of junk mail, and even cleans our floors. At IKIN, AI is used to realize convincing dimensional holograms in ambient light. AI-assisted head-tracking technology establishes convincing perspectives, replacing the much more common headgear approach to 3D XR. And it is used by developers to easily convert 2D content into holographic 3D and to create original holographic imagery.
We may not be at the point of Self-Aware AI just yet, but as the technology continues to expand in use and refine its forms, the creation of a sentient AI may be less a matter of if than of when.
Read More On The IKIN Blog
To learn more about IKIN’s ongoing developments in holographic technology and hologram applications, please visit the IKIN Blog. And, remember to follow IKIN on social media.
