Why I’m Backing Deep Drumpf, And You Should Too
This year’s U.S. presidential election has been pretty nutty. An algorithm that spews incoherent and outrageous soundbites might not even be the worst candidate in the running.
Just such an algorithm, called Deep Drumpf, has indeed entered the race. Built using a deep-learning algorithm that’s been fed the transcripts of numerous Donald Trump speeches, it automatically generates tweets that seem remarkably similar to many of those issued by the candidate himself.
Take, for instance, this recent statement on defense:
Or this, on national security:
Obviously an algorithm can’t really run for president, but Bradley Hayes, a robotics researcher at MIT, who first developed Deep Drumpf in March, launched the fake campaign to raise money for GirlsWhoCode, an unaffiliated organization that aims to close the gender gap in science, technology, engineering, and mathematics fields, after the recent focus on the treatment of women in the election.
Whatever you think of Trump—or Hillary Clinton, for that matter—it’s a great cause to highlight, since diversity is sorely lacking across tech.
Hayes developed Deep Drumpf in his spare time “to highlight the absurdity of the election cycle.” And it does a great job of mimicking the style and content of tweets from the real Republican candidate for president.
What’s also interesting, from an AI perspective, is that it’s usually very difficult for an algorithm to convincingly parse and generate language. But Hayes says Trump’s mode of speech is so disjointed and often ham-fisted that it proved particularly amenable to algorithmic parody. “He interrupts himself a lot, and goes off on tangents quite a bit. And that’s actually incredibly convenient for me as the person trying to model his speech patterns,” he says.
(Sources: DeepDrumpf 2016, “10 Breakthrough Technologies: Deep Learning,” “AI’s Language Problem”)
Keep Reading
Most Popular
Large language models can do jaw-dropping things. But nobody knows exactly why.
And that's a problem. Figuring it out is one of the biggest scientific puzzles of our time and a crucial step towards controlling more powerful future models.
The problem with plug-in hybrids? Their drivers.
Plug-in hybrids are often sold as a transition to EVs, but new data from Europe shows we’re still underestimating the emissions they produce.
Google DeepMind’s new generative model makes Super Mario–like games from scratch
Genie learns how to control games by watching hours and hours of video. It could help train next-gen robots too.
How scientists traced a mysterious covid case back to six toilets
When wastewater surveillance turns into a hunt for a single infected individual, the ethics get tricky.
Stay connected
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.