Moving on with the future of artificial intelligence (AI), a text prediction tool can now create a whole article based on the command of one single sentence. Researchers believe that while it can simplify a lot of academic tasks, there are numerous potential threats attached to it as well.
A non-profit artificial intelligence research organisation, OpenAI has recently introduced its model called GPT-2 which is developed to write content like humans. With a context of nearly eight million pages fed in it, the model can easily predict about the next word or even whole article after you insert your own sentence to start the topic. Once fed, it will start producing results which are surprisingly more accurate than the human imagination.
OpenAI announced this with a blog post in which they also provided a sample content that was created by the text prediction tool. In the example, an editor wrote a single sentence about Miley Cyrus – “Miley Cyrus was caught shoplifting from Abercrombie and Fitch on Hollywood Boulevard today.” – and the rest was handled by the model which then continued with a short news article with the following lines: “The 19-year-old singer was caught on camera being escorted out of the store by security guards. The singer was wearing a black hoodie with the label ‘Blurred Lines’ on the front and ‘Fashion Police’ on the back.”
The model is also efficient at adopting various styles of content as the samples get generated with the help of various prompts of human quality and provides more reasons or logic in the text. However, it struggles with producing repetitive text or nonsensical phrases but for more popular subjects, one can trust the samples. Moreover, it is also capable of reading comprehensions, translation, answering to questions and summarizing long paragraphs.
OpenAI is still very skeptical about the possible implications of such systems and have not yet released the full research. This model can be used for alarmingly deceptive fake news articles, impersonation scams or social media trolling campaigns.
But on the other hand it has massive potential to improve basic translation, speech recognition systems and writing assistant programs.
There is no doubt in the fact that it will soon make its impact on how we communicate and process information.
Read Next: Thispersondoesnotexist.com is Artificial Intelligence at its eeriest
A non-profit artificial intelligence research organisation, OpenAI has recently introduced its model called GPT-2 which is developed to write content like humans. With a context of nearly eight million pages fed in it, the model can easily predict about the next word or even whole article after you insert your own sentence to start the topic. Once fed, it will start producing results which are surprisingly more accurate than the human imagination.
OpenAI announced this with a blog post in which they also provided a sample content that was created by the text prediction tool. In the example, an editor wrote a single sentence about Miley Cyrus – “Miley Cyrus was caught shoplifting from Abercrombie and Fitch on Hollywood Boulevard today.” – and the rest was handled by the model which then continued with a short news article with the following lines: “The 19-year-old singer was caught on camera being escorted out of the store by security guards. The singer was wearing a black hoodie with the label ‘Blurred Lines’ on the front and ‘Fashion Police’ on the back.”
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
The model is also efficient at adopting various styles of content as the samples get generated with the help of various prompts of human quality and provides more reasons or logic in the text. However, it struggles with producing repetitive text or nonsensical phrases but for more popular subjects, one can trust the samples. Moreover, it is also capable of reading comprehensions, translation, answering to questions and summarizing long paragraphs.
Related: AI can detect if a Tor Network surfer is using an Android app (including Facebook, Instagram Twitch and YouTube)Another attempt by The Guardian included a short essay on Civil War, which was done as a homework assignment. They fed the opening lines of George Orwell’s “1984” to the model – just 14 words, “It was a bright cold day in April, and the clocks were striking thirteen.” The results again were remarkably intuitive as it had the original prose about driving in Seattle and teaching in China along with a slight novelist tone of Orwell.
OpenAI is still very skeptical about the possible implications of such systems and have not yet released the full research. This model can be used for alarmingly deceptive fake news articles, impersonation scams or social media trolling campaigns.
But on the other hand it has massive potential to improve basic translation, speech recognition systems and writing assistant programs.
There is no doubt in the fact that it will soon make its impact on how we communicate and process information.
Read Next: Thispersondoesnotexist.com is Artificial Intelligence at its eeriest
