The invention was the AI sensation of the 18th-century Habsburg court. Described as an “intelligent automaton,” it was claimed to be a machine able to play chess on its own.
The device wowed Empress Maria Theresa and onlookers. It beat high-profile opponents and was put on display in city after city. Skeptics invited to examine the machine could see no trick. Believers were seduced by stereotypes about eastern mysticism, even sorcery.
The problem is that the skeptics were right. The (unfortunately named) Mechanical Turk was controlled by a person cunningly hidden within its bowels. The secret was not published until years after the machine was destroyed in a fire.
This story comes to mind when considering some more recent news from the world of artificial intelligence. Too often, legitimate advances in this field are overshadowed by boosterism and misdirection.
This is part of a bigger pattern, in which both backers and critics of AI make extreme claims. Often, the potential upsides and downsides are overstated. The reality is that, while there are likely to be big changes, we don’t know on what scale or timeline.
Andrew Coyne: The AI job apocalypse is coming. Or is it?
Only last week, a senior safety researcher at the AI company Anthropic resigned, saying he was worried about the future of the world and warning that the organization found it hard to live up to its values.
Given all the uncertainty over how the AI future will unfold, it’s important to approach these debates with a large pinch of salt. Not all is necessarily as it appears.
This was evident in a big way when Builder.ai, which had been valued at US$1.5-billion, declared bankruptcy last year, burning investors. The company purported to be using AI to build apps but had actually relied on people in India to do the work. A modern-day Mechanical Turk.
Also getting assistance are the supposedly autonomous vehicles operated by the company Waymo. According to an executive at a Senate hearing this month in the United States, it turns out that these have a sort of phone-a-friend option. When the vehicles run into a problem the machine can’t solve, they are told what to do by a person in the Philippines.
The problem is that the company’s website states plainly, with no caveats or qualifiers, that “Our Waymo vehicles are fully autonomous.” Although the company operates taxis, anything feeding the general perception that autonomous driving technology is more advanced than it really is is dangerous for road safety. The current level of driver assistance technology cannot replace the need to be awake, sober and paying attention.
Another example of humans lurking behind the curtain was at the Reddit-like platform Moltbook. This was conceived as a social media network for AI bots, allowing them to communicate with each other.
A few weeks ago, there was a flurry of news stories as people noticed that bots here were showing signs of becoming self-aware. Some of them even seemed to be conspiring against humans. This suggested a step toward technological “singularity,” the point at which machines slip the bonds of human control.
Happily, experts concluded that the bots’ behaviour was more benign: they were regurgitating concepts from the material used to train them, which was apparently heavy on science fiction. The good news is that the terminators haven’t moved a step closer to reality. But the bad news is that so many people were so credulous, so ready to believe the hype.
A recent New Yorker article detailed the ways in which AI is both improving greatly and still profoundly flawed. The reporter described efforts by the company Anthropic to stress-test its chatbot, Claude. The results were sometimes ridiculous – the chatbot, which is only lines of code, believed itself capable of going to a meeting in person – but also showed Claude fumbling toward something that might be called intelligence.
The loudest proselytizers for artificial intelligence like to present the technology as inevitable. That may be so. However, we all need to be sensible enough not to fall for tech-bro bravado. Exaggerating what AI has actually achieved cannot be a way to convince people that this future has already arrived.
For what it’s worth, a sort of Mechanical Turk was, eventually, invented. In 1997 a computer named Deep Blue became the first to beat a reigning chess champion. This was two centuries after the device that so fascinated the Empress.