Where many firms are doing endless efforts to combat fake news, an Elon Musk-backed company has developed a new artificially intelligent (AI) system that can write seriously plausible-looking fake news.
Musk’s OpenAI has created an algorithm that generates believable-looking fake news stories on any topic just with the help of some keywords as a starting point.
The AI algorithm named GPT-2 was initially designed as a generalized language AI that can answer questions, summarize stories and translate text. However, the team realized that the tool could be used for more sinister purposes such as spreading of convincing false information, social-media posts or other text content on large scale.
We've trained an unsupervised language model that can generate coherent paragraphs and perform rudimentary reading comprehension, machine translation, question answering, and summarization — all without task-specific training: https://t.co/sY30aQM7hU pic.twitter.com/360bGgoea3
— OpenAI (@OpenAI) February 14, 2019
GPT-2 was trained on around eight million web pages. Given a prompt, the AI is tasked with predicting the next word based on how those words have been used on the websites it read. In the end, the algorithm churns out passages of text that are way more coherent than past attempts to build AI with contextual knowledge of language, reported Futurism.
However, due to its ability to spread disinformation to masses, OpenAI has decided to keep the algorithm closed and not open for the public since it can easily be used to crate misleading news articles, impersonate people or do other shady things.
As a result, according to MIT Technology Review, the team plans to make a ‘simplified version’ of its AI available to public. “It could be that someone who has malicious intent would be able to generate high quality fake news,” David Luan, OpenAI’s vice president of engineering told Wired.