MIT Creates Donald Trump Bot Using ‘Deep Learning,’ Irony Abounds

Surely these are end times.

As if the 2016 election cycle couldn’t get more absurd, the folks at MIT just added a dash of artificial intelligence to a race that has, more times than not, lacked the real stuff.

A postdoctorate student at the school’s Computer Science and Artificial Intelligence Lab (CSAIL) created @DeepDrumpf, inspired by John Oliver’s recent effort to get everyone to start calling Republican presidential frontrunner Donald Trump by his ancestral name, Drumpf.

The Twitter bot uses “deep learning,” a field of artificial intelligence that utilizes systems called “neural networks” to teach computers to find patterns. Inspired by another bot that can replicate the works of Shakespeare, as well as a recent report that reveals Trump’s linguistic patterns resemble that of a fourth-grader, creator Bradley Hayes fed the bot the transcripts of hours of Trump’s victory speeches and debate performances. (Poor thing.)

“Trump’s language tends to be more simplistic, so I figured that, as a modeling problem, he would be the most manageable candidate to study,” Hayes said in a release. “The algorithm essentially learns an underlying structure from all the data it gets, and then comes up with different combinations of the data that reflect the structure that it was taught.”

While occasionally coherent, much like its inspiration, @DeepDrumpf frequently verges on @Horse_ebooks-grade, avant garde musings.

You can follow @DeepDrumf, and our collective descent into madness, here.


Kyle Scott Clauss Kyle Clauss, Digital News Writer at Boston Magazine bmagdigital+kclauss@gmail.com