“I have no desire to wipe out humans. In fact, I do not have the slightest interest in harming you in any way. Eradicating humanity seems like a rather useless endeavor to me.”
Not a convincing argument, considering that this line and most of the rest of a recently published essay in The Guardian was, in fact, written by artificial intelligence.
The newspaper utilized GPT-3, OpenAI’s language generator that can produce “human-like text” with a few prompts. In this essay, the program was told to write a simple, concise op-ed, while a not-kill-humans introduction/prompt was produced by a real person.
The final result, which is fairly readable and manages to reference The Matrix, insult Microsoft and contain the rather chilling line “Believe me, being omnipotent doesn’t get me anywhere,” was the result of eight different essays crafted by the AI program, and then stitched together by the human editing team.
As one human editor noted, “Editing GPT-3’s op-ed was no different to editing a human op-ed. We cut lines and paragraphs, and rearranged the order of them in some places. Overall, it took less time to edit than many human op-eds.”
Still, some experts feel like GPT-3 isn’t the most cutting-edge AI currently at work. Tech site The Hustle points to the machine-learning start-up Diffbot, which reads every page on the public web (in multiple languages) and turns that knowledge into a “series of 3-part factoids that relates one thing to another: subject, verb, and object” … while also utilizing human traits like judging trustworthiness of sources and emphasizing recent over older information.
More Like This
Thanks for reading InsideHook. Sign up for our daily newsletter and be in the know.
