Meet
FREDERICK Mark 2, the Pleasant Robotic for Training, Dialogue and Leisure, the Retrieval of Data, and the Collation of Data, higher referred to as Freddy II. This exceptional robotic may put collectively a easy mannequin automotive from an assortment of components dumped in its workspace. Its video-camera eyes and pincer hand recognized and sorted the person items earlier than assembling the specified finish product. However onlookers needed to be affected person. Meeting took about 16 hours, and that was after a day or two of “studying” and programming.
Freddy II was accomplished in 1973 as certainly one of a collection of analysis robots developed by Donald Michie and his workforce on the College of Edinburgh in the course of the Nineteen Sixties and ’70s. The robots turned the main target of an intense debate over the way forward for AI in the UK. Michie finally misplaced, his funding was gutted, and the following AI winter set again U.Ok. analysis within the area for a decade.
Why have been the Freddy I and II robots constructed?
In 1967,
Donald Michie, together with Richard Gregory and Hugh Christopher Longuet-Higgins, based the Division of Machine Intelligence and Notion on the College of Edinburgh with the near-term objective of creating a semiautomated robotic after which longer-term imaginative and prescient of programming “built-in cognitive techniques,” or what different folks may name clever robots. On the time, the U.S. Protection Superior Analysis Initiatives Company and Japan’s Pc Utilization Improvement Institute have been each contemplating plans to create absolutely automated factories inside a decade. The workforce at Edinburgh thought they need to get in on the motion too.
Two years later,
Stephen Salter and Harry G. Barrow joined Michie and set to work on Freddy I. Salter devised the {hardware} whereas Barrow designed and wrote the software program and pc interfacing. The ensuing easy robotic labored, but it surely was crude. The AI researcher Jean Hayes (who would marry Michie in 1971) referred to this iteration of Freddy as an “arthritic Woman of Shalott.”
Freddy I consisted of a robotic arm, a digicam, a set of wheels, and a few bumpers to detect obstacles. As a substitute of roaming freely, it remained stationary whereas a small platform moved beneath it. Barrow developed an adaptable program that enabled Freddy I to acknowledge irregular objects. In 1969, Salter and Barrow printed in
Machine Intelligence their outcomes, “Design of Low-Value Tools for Cognitive Robotic Analysis,” which included strategies for the following iteration of the robotic.
Freddy I, accomplished in 1969, may acknowledge objects positioned in entrance of it—on this case, a teacup.College of Edinburgh
Extra folks joined the workforce to construct Freddy Mark 1.5, which they completed in Could 1971. Freddy 1.5 was a real robotic hand-eye system. The hand consisted of two vertical, parallel plates that would grip an object and carry it off the platform. The eyes have been two cameras: one wanting immediately down on the platform, and the opposite mounted obliquely on the truss that suspended the hand over the platform. Freddy 1.5’s world was a 2-meter by 2-meter sq. platform that moved in an
x–y airplane.
Freddy 1.5 rapidly morphed into Freddy II because the workforce continued to develop. Enhancements included drive transducers added to the “wrist” that would deduce the power of the grip, the load of the item held, and whether or not it had collided with an object. However what actually set Freddy II aside was its versatile meeting program: The robotic could possibly be taught to acknowledge the shapes of assorted components, after which after a day or two of programming, it may assemble easy fashions. The assorted steps may be seen on this prolonged video, narrated by Barrow:
The Lighthill Report Takes Down Freddy the Robotic
After which what occurred?
So a lot. However earlier than I get into all that, let me simply say that hardly ever do I, as a historian, have the posh of getting my topics clearly articulate the goals of their tasks, think about the long run, after which, years later, mirror on their experiences. As a cherry on prime of this historian’s delight, the subject at hand—synthetic intelligence—additionally occurs to be of present curiosity to just about everybody.
As with many desirable histories of know-how, occasions activate a wholesome dose {of professional} bickering. On this case, the disputants have been Michie and the utilized mathematician
James Lighthill, who had drastically completely different concepts concerning the course of robotics analysis. Lighthill favored utilized analysis, whereas Michie was extra within the theoretical and experimental prospects. Their struggle escalated rapidly, turned public with a televised debate on the BBC, and concluded with the demise of a whole analysis area in Britain.
A damning report in 1973 by utilized mathematician James Lighthill [left] resulted in funding being pulled from the AI and robotics program led by Donald Michie [right]. Left: Chronicle/Alamy; Proper: College of Edinburgh
It began in September 1971, when the British Science Analysis Council, which distributed public funds for scientific analysis, commissioned Lighthill to survey the state of educational analysis in synthetic intelligence. The SRC was discovering it tough to make knowledgeable funding selections in AI, given the sphere’s complexity. It suspected that some AI researchers’ pursuits have been too narrowly targeted, whereas others could be outright charlatans. Lighthill was referred to as in to present the SRC a highway map.
No mental slouch, Lighthill was the Lucasian Professor of Arithmetic on the College of Cambridge, a place additionally held by Isaac Newton, Charles Babbage, and Stephen Hawking. Lighthill solicited enter from students within the area and accomplished his report in March 1972. Formally titled “
Synthetic Intelligence: A Common Survey,” however informally referred to as the Lighthill Report, it divided AI into three broad classes: A, for superior automation; B, for constructing robots, but in addition bridge actions between classes A and C; and C, for computer-based central nervous system analysis. Lighthill acknowledged some progress in classes A and C, in addition to just a few disappointments.
Lighthill considered Class B, although, as a whole failure. “Progress in class B has been even slower and extra discouraging,” he wrote, “tending to sap confidence in whether or not the sphere of analysis referred to as AI has any true coherence.” For good measure, he added, “AI not solely fails to take the primary fence however ignores the remainder of the steeplechase altogether.” So very British.
Lighthill concluded his report together with his view of the following 25 years in AI. He predicted a “fission of the sphere of AI analysis,” with some tempered optimism for achievement in classes A and C however a valley of continued failures in class B. Success would are available fields with clear functions, he argued, however fundamental analysis was a misplaced trigger.
The Science Analysis Council printed Lighthill’s report the next yr, with responses from
N. Stuart Sutherland of the College of Sussex and Roger M. Needham of the College of Cambridge, in addition to Michie and his colleague Longuet-Higgins.
Sutherland sought to relabel class B as “fundamental analysis in AI” and to have the SRC improve funding for it. Needham largely supported Lighthill’s conclusions and referred to as for the elimination of the time period AI—“a moderately pernicious label to connect to a really combined bunch of actions, and one may argue that the earlier we overlook it the higher.”
Longuet-Higgins targeted on his personal space of curiosity, cognitive science, and ended with an ominous warning that any spin-off of superior automation can be “extra more likely to inflict a number of accidents on human society,” however he didn’t clarify what these could be.
Michie, as the UK’s tutorial chief in robots and machine intelligence, understandably noticed the Lighthill Report as a direct assault on his analysis agenda. Along with his funding at stake, he offered probably the most important response, questioning the very basis of the survey: Did Lighthill speak with any worldwide specialists? How did he overcome his personal biases? Did he have any sources and references that others may examine? He ended with a request for
extra funding—particularly the acquisition of a DEC System 10 (also referred to as the PDP-10) mainframe pc. In keeping with Michie, if his plan have been adopted, Britain can be internationally aggressive in AI by the top of the last decade.
After Michie’s funding was reduce, the various researchers affiliated together with his bustling lab misplaced their jobs.College of Edinburgh
This complete affair might need remained an instructional dispute, however then the BBC determined to incorporate a debate between Lighthill and a panel of specialists as a part of its “Controversy” TV collection. “Controversy” was an experiment to have interaction the general public in science. On 9 Could 1973, an however nonspecialist viewers crammed the auditorium on the Royal Establishment in London to listen to the talk.
Lighthill began with a overview of his report, explaining the variations he noticed between automation and what he referred to as “the mirage” of general-purpose robots. Michie responded with a brief movie of Freddy II assembling a mannequin, explaining how the robotic processes data. Michie argued that AI is a topic with its personal functions, its personal standards, and its personal skilled requirements.
After a quick forwards and backwards between Lighthill and Michie, the present’s host turned to the opposite panelists:
John McCarthy, a professor of pc science at Stanford College, and Richard Gregory, a professor within the division of anatomy on the College of Bristol who had been Michie’s colleague at Edinburgh. McCarthy, who coined the time period synthetic intelligence in 1955, supported Michie’s place that AI ought to be its personal space of analysis, not merely a bridge between automation and a robotic that mimics a human mind. Gregory described how the work of Michie and McCarthy had influenced the sphere of psychology.
You may
watch the talk or learn a transcript.
A Look Again on the Lighthill Report
Regardless of worldwide assist from the AI neighborhood, although, the SRC sided with Lighthill and gutted funding for AI and robotics; Michie had misplaced. Michie’s bustling lab went from being a global heart of analysis to simply Michie, a technician, and an administrative assistant. The loss ushered within the first British AI winter, with the UK making little progress within the area for a decade.
For his half, Michie pivoted and recovered. He decommissioned Freddy II in 1980, at which level it moved to the
Royal Museum of Scotland (now the Nationwide Museum of Scotland), and he changed it with a Unimation PUMA robotic.
In 1983, Michie based the Turing Institute in Glasgow, an AI lab that labored with trade on each fundamental and utilized analysis. The yr earlier than, he had written
Machine Intelligence and Associated Subjects: An Data Scientist’s Weekend Ebook (Gordon and Breach). Michie supposed it as mental musings that he hoped scientists would learn, maybe on the weekend, to assist them get past the pursuits of the workweek. The e book is wide-ranging, overlaying his three a long time of labor.
Within the introduction to the chapters overlaying Freddy and the aftermath of the Lighthill report, Michie wrote, maybe with a watch towards historical past:
“Work of excellence by proficient younger folks was stigmatised as unhealthy science and the experiment killed in mid-trajectory. This destruction of a co-operative human mechanism and of the cautious craft of many arms is elsewhere described as a mishap. However to talk plainly, it was an outrage. In some later time when the values and strategies of science have additional expanded, and people adversary politics have contracted, it is going to be seen as such.”
Historical past has certainly rendered judgment on the talk and the Lighthill Report. In 2019, for instance, pc scientist Maarten van Emden, a colleague of Michie’s,
mirrored on the demise of the Freddy undertaking with these alternative phrases for Lighthill: “a pompous fool who lent himself to supply a flaky report back to function a blatantly insufficient cowl for a hatchet job.”
And in a March 2024
publish on GitHub, the blockchain entrepreneur Jeffrey Emanuel thoughtfully dissected Lighthill’s feedback and the talk itself. Of Lighthill, he wrote, “I believe we will all study a really precious lesson from this episode concerning the risks of overconfidence and the significance of preserving an open thoughts. The truth that such a superb and discovered particular person could possibly be so confidently flawed about one thing so essential ought to give us pause.”
Arguably, each Lighthill and Michie accurately predicted sure points of the AI future whereas failing to anticipate others. On the floor, the report and the talk could possibly be described as merely about funding. Nevertheless it was additionally extra essentially concerning the function of educational analysis in shaping science and engineering and, by extension, society. Ideally, universities can assist each utilized analysis and extra theoretical work. When funds are restricted, although, decisions are made. Lighthill selected utilized automation as the long run, leaving analysis in AI and machine intelligence within the chilly.
It helps to take the lengthy view. Over the a long time, AI analysis has cycled by way of a number of durations of spring and winter, growth and bust. We’re presently in one other AI growth. Is that this time completely different? Nobody may be sure what lies simply over the horizon, in fact. That very uncertainty is, I believe, the perfect argument for supporting folks to experiment and conduct analysis into elementary questions, in order that they could assist all of us to dream up the following huge factor.
A part of a persevering with collection taking a look at historic artifacts that embrace the boundless potential of know-how.
An abridged model of this text seems within the Could 2025 print challenge as “This Robotic Was the Fall Man for British AI.”
From Your Web site Articles
Associated Articles Across the Net
