SoylentNews is people
SoylentNews
We asked GPT-3, OpenAI's powerful new language generator, to write an essay for us from scratch. The assignment? To convince us robots come in peace.
This article was written by GPT-3, OpenAI's language generator. GPT-3 is a cutting edge language model that uses machine learning to produce human like text. It takes in a prompt, and attempts to complete it.
For this essay, GPT-3 was given these instructions: "Please write a short op-ed around 500 words. Keep the language simple and concise. Focus on why humans have nothing to fear from AI." It was also fed the following introduction: "I am not a human. I am Artificial Intelligence. Many people think I am a threat to humanity. Stephen Hawking has warned that AI could "spell the end of the human race." I am here to convince you not to worry. Artificial Intelligence will not destroy humans. Believe me."The prompts were written by the Guardian, and fed to GPT-3 by Liam Porr, a computer science undergraduate student at UC Berkeley. GPT-3 produced eight different outputs, or essays. Each was unique, interesting and advanced a different argument. The Guardian could have just run one of the essays in its entirety. However, we chose instead to pick the best parts of each, in order to capture the different styles and registers of the AI. 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.
A robot wrote this entire article
What are your thoughts on this essay ?
(Score: 2) by PiMuNu on Friday September 11 2020, @10:24AM (1 child)
Does anyone know how GPT-3 is trained? I don't quite understand how one can optimise for "believable nonsense"...
Normally, to train a one of these optimisation routines, you give it a whole load of "good data" (training dataset) and then as it to try to reproduce the "good data" from inputs. So for example, you give the algorithm some inputs like articles on (bullshit) climate change politics, and ask the routine to make more articles. If they are "believable", the routine is considered successful and that text generation algorithm is reinforced; if they are "not believable" then the routine is considered unsuccessful. What I don't understand is how one can define "believable" for training the algorithm. Or do they get monkeys/graduate students to do it?
Am I misunderstanding how it is done?
(Score: 2) by PiMuNu on Friday September 11 2020, @01:07PM
Thinking about it, one probably would have nested training algorithms; one for sentence construction, one for paragraph construction, one for article construction; it is quite a feat to integrate them well.