by Daniel Bork, Startup Shut-Downer
Tech stocks were roiled Monday when the Canadian Securities and Exchange Commission reported that the up-and-coming computer science subfield of “deep learning” was, in fact, a type of massively parallel financial fraud known as a pyramid scheme. Ringleader Geoffrey Hinton of the University of Toronto seemed to confess his involvement in an interview with federal prosecutors last month, the transcript of which was made public at his Monday court hearing.
“It’s all a great big lie,” said Hinton when pressed to explain inconsistent results claimed by the alleged breakthrough. “It was just a few MATLAB scripts and a lot of compiler tricks. I’m finished.” Hinton went on to detail the byzantine financial shell game that he and his accountants used to mislead backers and regulatory agencies, in which so-called “hidden layers” of investors recruiting ever-growing numbers of new contributors into the scheme. “Sorry to burst your bubble, but that’s how it is in Silicon Valley,” claimed Hinton. “Everyone thinks that somewhere in the chain of startups, corporations, and angel investors is a real technology that actually does something, but nope. It’s just suckers all the way down.”
According to the investigation’s report, many promising “deep learning” applications that have received popular press, such as Google’s “DeepMind” software, may actually be little more than flashy demonstrations designed to attract investors while papering over the lack of any real technical achievement. “Yeah, it can make pretty pictures,” Hinton confessed, “and maybe beat a few people at board games, but that’s about it. It’s just a buzzword. It doesn’t solve any real problems for anyone.”
“You can’t just tack the word ‘deep’ onto any phrase and expect that to somehow make it more profound,” added Hinton, “and, obviously, machines can’t learn. Come on, people.”
Reactions across the tech world were mixed, ranging from horrified apocalyptic pronouncements to more sedate, reassuring apocalyptic pronouncements. PayPal cofounder Peter Thiel, for instance, comforted an audience of troubled graduate students at Berkeley’s Machine Intelligence Research Institute by stating that, while “the glorious dawning age of the Singularity may dawn a bit later than expected,” he nevertheless expected that “the generous contributions of our friends at the Trump Administration will keep every one of your projects funded bigly (sic) for decades to come”. The White House did not respond to requests for comment.