icon bookmark-bicon bookmarkicon cameraicon checkicon chevron downicon chevron lefticon chevron righticon chevron upicon closeicon v-compressicon downloadicon editicon v-expandicon fbicon fileicon filtericon flag ruicon full chevron downicon full chevron lefticon full chevron righticon full chevron upicon gpicon insicon mailicon moveicon-musicicon mutedicon nomutedicon okicon v-pauseicon v-playicon searchicon shareicon sign inicon sign upicon stepbackicon stepforicon swipe downicon tagicon tagsicon tgicon trashicon twicon vkicon yticon wticon fm
27 Jan, 2016 17:01

Move over, Abe Lincoln? Now an AI system can write halfway-decent political speeches

Move over, Abe Lincoln? Now an AI system can write halfway-decent political speeches

Forget Lincoln and Churchill: Now even a machine can produce a convincing political speech, and this may be bad news for human speechwriters. Researchers have found a way to get AI programs to write speeches by analyzing thousands of previous ones.

Researchers at the University of Massachusetts, Amherst built a machine that can learn from 4,000 segments of political speeches from 53 US Congressional floor debates.

“In this report we present a system that can generate political speeches for a desired political party,” researcher Valentin Kassarnig wrote in a recent publication.

“Furthermore, the system allows to specify whether a speech should hold a supportive or opposing opinion.”

After trying various techniques to analyze the data, he decided to use “n-grams” – an approach made famous by Google studying historical word frequecy – that looks for sequences of “n” words of phrases to make sense of a text. It analyzes parts of speech and tags words based on their grammatical role, like verb, noun or adjective.

The program used this method to sift through 50,000 sentences, each with an average of 23 words. Every sentence was tagged with a political allegiance, specifying whether the one giving the speech was endorsing or opposing the subject that they were talking about. Speeches were further divided by the team based on whether the person making it was a Democrat or a Republican.

The machine produces original text using just a random five-word seed of a speech and then putting one word after another by finding the words that has the highest probability of appearing there based on the topic. From there, the entire speech is automatically constructed.

“That allows us to determine very quickly all words which can occur after the previous five ones and how likely each of them is,” Kassarnig said.

READ MORE: Apple acquires artificial intelligence start-up that reads facial emotions

Here’s an example of what the AI is capable of:

“Mr. Speaker, for years, honest but unfortunate consumers have had the ability to plead their case to come under bankruptcy protection and have their reasonable and valid debts discharged. The way the system is supposed to work, the bankruptcy court evaluates various factors including income, assets and debt to determine what debts can be paid and how consumers can get back on their feet. Stand up for growth and opportunity. Pass this legislation.”

Well, Gettysburg Address it ain’t. But it’s getting there.

Podcasts
0:00
25:59
0:00
26:57