In February, an artificial intelligence lab, which was co-founded by Elon Musk, revealed to the world that its latest breakthrough was too risky to launch for the public. OpenAI claimed it had made the language software so fluent at generating text that it might be acquired to crank out spam or fake news.
On Thursday, two recent masters graduates in computer science released something that they say is a recreation of OpenAI’s withheld software onto the internet for anyone to download and utilize.
Aaron Gokaslan, 23, and Vanya Cohen, 24, say they aren’t out to cause havoc and don’t believe such software poses many risks to the society yet. The duo says their release was intended to show that one need not be an elite lab who’s rich in dollars and possess PhDs to create this kind of software: They used an estimated $50,000 worth of cloud computing from Google for free, which hands out credits to academic institutions. And they argue whether setting their creation free, which can help others explore and prepare for future advances, is a boon or a bane.
“This allows everyone to have an important conversation about security and researchers to help secure against future potential abuses,” says Cohen, who notes language software also has many positive uses. “I’ve gotten scores of messages and most of them have been like ‘Way to go.'”
The pair’s experiment, like OpenAI’s, included providing machine learning software text from millions of webpages collected by harvesting links shared on Reddit. Once the software internalizes patterns of language from the provided text, it can then be adapted to multiple tasks such as translation, powering chatbots, or generating new text in response to a prompt.