IBM has been pioneering AI systems since before the technology had fully escaped the realm of science fiction
Long before there were smart speakers, language translation apps and game-playing bots, IBM was laying the groundwork for AI-powered functionality across business and society.
As early as the 1920s, the company was making forays into automated translation. Its first system combined human agents with the era’s latest technologies — headphones and a switchboard. Within a few decades the concept evolved into an experimental software program that automatically translated Russian to English on the 701 Electronic Data Processing Machine, the world’s first commercial scientific computer. Today, the IBM Watson Language Translator, powered by neural machine translation technology, can instantly translate documents, news and more into dozens of languages.
Natural language processing (NLP) has traveled a similar path. Today’s smart speakers and automated customer service agents owe a debt to the development of IBM’s experimental speech-recognition device, Shoebox, which debuted at the 1962 World’s Fair. Named for its size, Shoebox entranced crowds by calculating a series of numbers and mathematical commands that had been spoken to it. It was an early hint of today’s NLP systems, which boost our productivity, entertain us and help us navigate the world.
Much of IBM’s more recent progress in AI comes from a willingness to have a bit of fun. Rules-based board and strategy games enable AI systems to study and anticipate interactions. Honing such skills leads to better game play, but also garners insight for researchers into how best to train AI systems across a variety of tasks.
In 1992, IBM created the world’s first backgammon-playing computer, which used an early neural network to learn game play and strategy. It had brainpower equivalent to a sea slug. Within a half decade, the chess-playing supercomputer Deep Blue became the first machine to defeat a world champion.
In 2011, Watson beat two of the most successful Jeopardy! champions. These systems have become the basis for AI advances that help companies and municipalities around the world manage factories, supply chains and urban infrastructure, and create more efficient business processes, from hiring and retention to lending practices.
As AI continues to extend its reach, IBM has committed to being a leader in creating more transparent AI systems that reduce unintended bias, as well as in outlining safeguards to maintain privacy in an age of ubiquitous facial recognition.
Read on to learn more about the significant role IBM has played over the years in developing the AI systems that so many of us rely on every day — as well as the work that the company continues to do to improve AI technologies and policies.
These are the stories of the company that harnessed the passion and talents of its employees to help turn science fiction into reality. This is IBM in space.
IBM’s impact on space exploration is nearly impossible to overstate. In the 1940s, when spaceflight was still the realm of science fiction, IBM was piecing together the tools and expertise that would one day help launch NASA’s rockets into the cosmos. One of those tools, the IBM 701 Electronic Data Processing Machine, a large-scale electromagnetic calculator originally deployed to map ballistic trajectories, provided the company with the backbone to handle the government’s satellite programs in the 1950s.
By the time NASA began putting astronauts into Earth’s orbit with its Mercury and Gemini projects in the 1960s, IBM was providing cutting-edge data processing systems, complex software programs, and compact onboard computers to manage spacecraft safely from launch to landing. The company’s role in the later Apollo missions was of such importance, in fact, that one NASA flight director said, “Without IBM and the systems they provided, we would not have landed on the moon.”
By the time the lunar chapter came to a close, IBM was already helping to manifest the next leap forward for mankind. In the 1970s, NASA brought on the company to design the support systems for the country’s first space station, Skylab, as well as for the first international space mission, Apollo-Soyuz, a collaboration between the US and the former Soviet Union.
During the 1980s, IBM was consumed with creating the software and computers that would operate the Space Shuttle program.
At the dawn of another space age, one increasingly defined by entrepreneurialism and privatization, IBM was still playing a fundamental role in enabling cosmic exploration. The company continued to leverage emerging technologies to improve life on Earth and to reach for the stars.
In 2018, in a collaboration with Airbus, IBM introduced the first free-flying AI assistant in space — a bot named CIMON that provides companionship and critical technical support to the ISS’s European Columbus research module. Separately, the company has parked a space-based communications system in Earth’s lower orbit with the intention of mapping out the future of edge computing and cloud data centers. As Naeem Altaf, CTO of IBM’s Space Tech team said, “We are learning. We are redefining what is possible.”
A proud history of building the world’s most powerful computers
The first recorded use of the word “supercomputing” appeared in a 1929 edition of the New York World. The headline read “Super Computing Machines Shown,” referring to a desk-sized tabulator built by IBM for Columbia University. By today’s standards, the machine was hardly super — it could solve only 12 equations. But it inspired a focus at IBM that would underpin a century of building ever more powerful, innovative computers capable of tackling problems previously considered beyond the reach of any machine.
The first modern supercomputers — the term typically refers to mainframes designed to handle uncommonly demanding tasks in science or industry — came into use in the early 1960s when IBM introduced the IBM 7030 Stretch. It ranked as the fastest in the world for several years. In the decades to follow, IBM would continue to push supercomputing performance. Since 2000, IBM has consistently produced the most powerful machines. In 2018, IBM’s Summit and Sierra ranked as the world’s two fastest supercomputers — the only time a single vendor has achieved the top two positions.
IBM’s heavy outlays for research and a willingness to push the envelope of computer design has contributed to the company’s — and America’s — leadership in supercomputing. In a 1961 speech titled “Automation and National Power,” given shortly after the introduction of the IBM Stretch, IBM chairman and CEO Thomas J. Watson Jr. framed the pursuit of global dominance in supercomputing as important not only to IBM but to America’s economic competitiveness, national security and leadership in innovation, a mission that endures to this day. “We shall not hold back in the development of super machines,” Watson said. “The ‘faster than’ contest is not yet over. In some respects, Stretch is only the beginning in a new decade of super computers. Twenty years from now, we will surely look back on it as a relic of the pioneer days.”
The first supercomputers were built to meet the rigorous requirements of government laboratories engaged in nuclear weapons research, which often needed to conduct more than 100 billion arithmetical operations to evaluate a weapon’s design. Unlike traditional computers, supercomputers use more than one central processing unit (CPU). Thousands of processors work in parallel to perform calculations far faster than a standard computer, with its single CPU, can. By the 1970s, supercomputers began to make their way into the commercial mainstream, used by large financial institutions, manufacturers, insurers and retailers to crunch mountains of business and customer data faster than ever before.
The increase in supercomputing power has long followed Moore’s law, named after Intel co-founder Gordon Moore, who observed in 1965 that the processing power of computer chips was doubling roughly every two years. From the 1960s through the 1990s, supercomputers achieved roughly a thousandfold increase in speed per decade. In 2008, the IBM Roadrunnner supercomputer became the first to break through the petaflop barrier, capable of making more than 1 quadrillion floating-point calculations per second. (A quadrillion is a 1 followed by 15 zeros.)
Supercomputers have advanced to the point where they can simulate thinking. In 1997, IBM’s Deep Blue supercomputer defeated Garry Kasparov in a chess match, becoming the first computer system to defeat a reigning world chess champion.
Today, supercomputers are harnessed to design commercial aircraft, discover new drugs, devise more efficient batteries, discover new sources of energy, and provide insights into how diseases develop and identify promising treatments. The National Weather Service uses supercomputers made by IBM to process data from satellites, weather balloons, buoys and radar to produce up-to-the-minute warnings about severe weather.
Ultimately, supercomputing’s most profound impact has turned out to be improving everyday life in ways most of us are unaware.