Artificial intelligence (AI) can now write books, articles, lyrics, and whatever else you tell it to write. Its end result will make you question if it was written by a computer software or a human.
The software is called Generative Pre-training (GPT), an AI tool that generates on demand text that is eerily similar to a human’s composition. GPT runs on a neural network, a system designed to mirror the human brain through pattern recognition among large amounts of data.
The company that developed it, OpenAI, was originally a nonprofit doing AI research and founded by Elon Musk, Reid Hoffman and others. In 2018, Musk left the company but continued as a supporter and advisor. The following year, OpenAI turned for-profit and raised $1bn in funding from Microsoft, aiming to start commercializing its products.
So far, they have created three iterations of the GPT software, with GPT2 developed in 2020 but not released due to concerns about its potential misuse. Its latest version however, GPT3, is set to be its first product to become commercialized, with Reddit expressing interest.
For instance, the computer was given the opening sentence of George Orwell’s Nineteen Eighty-Four, which goes as follows:
“It was a bright cold day in April, and the clocks were striking thirteen.”
GPT3 then continued the passage based on this opening, inspired by the millions of literary samples it has read and processed. It was able to determine a particular style and tone, and adapted its output accordingly, writing:
“I was in my car on my way to a new job in Seattle. I put the gas in, put the key in, and then I let it run. I just imagined what the day would be like. A hundred years from now. In 2045, I was a teacher in some school in a poor part of rural China. I started with Chinese history and history of science.”
This represented a breakthrough moment for AI fans, but not only. Many have begun to imagine the multitude of ways this software could be used for day-to-day applications, generating both excitement and concern.
For example, this tool can enable companies that use chatbots to communicate more efficiently with their customers, providing a more personalized experience. It can be used for content generation, such as blog posts, articles or social media posts.
One of the most exciting uses of GPT technology is translation, as it can “learn” any language and generate text in that chosen language. This would make it easier for companies to reach new markets and international customers. As a highly efficient word processor, GPT3 could serve as a double check when translating books, although a human eye is still needed to correct minor errors.
But there are also some unsettling issues with this new technology.
The main concerns related to GPT3 are its potential uses for malicious intent, such as generating fake news or spamming people. Given the software’s computing power, it could quickly generate millions of articles that resemble the writing style of a human, making it nearly impossible for the average reader to discern between the two.
Moreover, GPT3 was also found to display all kinds of bias in its text, with its language often ending up being racist or sexist. This is a major flaw in any AI system that is calculating its output based on billions of parameters created by humans, whose work has historically displayed racial and gender discrimination.
