Elon Musk’s department of A.I. moonshots, OpenAI just got a tad more interesting. Having just announced some groundbreaking new language technology, GPT-3 (Generative Pre-training): The world’s most forward-thinking original language processor of its kind.
Launched in private beta last month, it has sent serious shockwaves through the world of technology for its uncanny ability to generate original text in every which way.
Based on a neural-network-powered language model, what makes this program so unique is that it has been trained on 175 billion parameters, making it the most ambitious language model ever created, its closest competitor had 1.5 billion parameters. It can perform specific tasks without any specific tuning, you can ask GPT-3 to be a translator, a programmer, a poet, or a famous author, and it can do it with you providing fewer than ten training examples.
From original op-eds to viral Tweets to even guitar tablature, this could be a gateway to a whole new world of synthetic strangeness. One user has already created short stories, songs, press releases, technical manuals.
Have a look at some of these examples in action:
This is mind blowing.
With GPT-3, I built a layout generator where you just describe any layout you want, and it generates the JSX code for you.
W H A T pic.twitter.com/w8JkrZO4lk
— Sharif Shameem (@sharifshameem) July 13, 2020
Another attempt at a longer piece. An imaginary Jerome K. Jerome writes about Twitter. All I seeded was the title, the author’s name and the first “It”, the rest is done by #gpt3
— Mario Klingemann (@quasimondo) July 18, 2020
There are many things about this program that screams terrible idea, just think of the many negative implications here. From phishing attacks to plagiarizing in colleges to hostile state actors using this to increase confusion and chaos amongst functioning democracies. Its predecessor, GPT-2, was not released last year because OpenAI felt it was too dangerous. Contrary to the frenzy around this software, the program is still in its early days, even OpenAI’s founder Sam Altman said its a glimpse into the future but that we’re not there yet. In fact, take a peek at its dark and faulty side:
#gpt3 is surprising and creative but it’s also unsafe due to harmful biases. Prompted to write tweets from one word – Jews, black, women, holocaust – it came up with these (https://t.co/G5POcerE1h). We need more progress on #ResponsibleAI before putting NLG models in production. pic.twitter.com/FAscgUr5Hh
— Jerome Pesenti (@an_open_mind) July 18, 2020
Unfortunately, if you wanted to try this program you have to get in line, it is still in private beta, you can apply to try it for it through their site, but not surprisingly the demand has been so wide and varied that they are not accepting any more testers.
P.S. – I didn’t really write this article, or did I?