Some suggested that the texts were written by a neural network, but other users put disadvantages to them, accusing them of being rude.
Student of the University of California at Berkeley Liam Porro for two weeks led blog Adolos , where he published motivational texts on productivity generated by the neural network GPT-3, which was developed by Elon Musk OpenAI. Almost no one understood that the posts were written by an algorithm, and those who guessed were criticized by users.
Porr wrote the headline and introduction, added photos, and the neural network wrote the rest of the text. In two weeks, the “blogger” attracted 26 thousand visitors, and 60 people even subscribed to it. The student noted that he was engaged in the promotion of the blog, but did not tell in what way.
The texts written by the neural network turned out to be coherent and relatively meaningful. In Porra’s posts, GPT-3 usually speaks to the reader in the first person and gives advice on how to become more productive , write better, or instill a reading habit .
You can’t be both lazy and a good writer at the same time. I’m sure some of you are thinking, “Of course you have to work hard! What is this advice? ” Well, I agree with you. In some ways.
I’ve noticed that when I sit down to write, I sometimes spend more time looking for the right words than writing. This is why I call the method “lazy writing.” It’s not about getting the job done. It’s about making the job as easy as possible.
I have nothing against hard work. Most of the time I enjoy it. But when I write, it seems to me that I work twice as hard, but with results twice as weak.
If you are like me – it is difficult for you to get started or you just want to improve what you are doing, then I hope this helps. Here are five tips for being lazy (but productive) while writing:
1. Forget about grammar.
2. Remove unnecessary words.
3. Write only when you want.
4. Let your ideas flow.
5. Ask someone to read what you have written.
from the article ” How to write more while thinking less “
The very first post “Feeling unproductive? Maybe you should stop thinking too much ”about the difference between“ creative thinking ”and“ reinforced thinking ”even made it to the top of the Hacker News forum.
In the comments, users began to complain about their unproductiveness and speculate that they really think too much. Some guessed that the text was written by a neural network, but their comments were minus.
Porr noted that only one person approached him personally with the question of whether the texts were created by a neural network.
After the experiment, the student suggested that GPT-3 could replace an entire team of content creators. At the same time, he admits that the quality of the texts is “doubtful” – the algorithm loses its logic, deviates from the topic, repeats the same thing. But with minimal editing, posts can be made “acceptable”. “Cut out the insignificant, add the conclusion and that’s it – people will not guess about the catch,” – says the author of the blog.
People say that GPT-3 is often incoherent and illogical. But this does not prevent them from reading and liking the texts.
Porr noted that GPT-3 can help big media companies keep the budget. He calculated that 133 BuzzFeed journalists would be able to produce the same amount of content using the neural network as 200 people without it. Thus, the algorithm would help save $ 3 million a year.
Porr emphasized that the purpose of the experiment was to show how the OpenAI algorithm can be used. “Perhaps some people will be angry with me for cheating them, but still,” the student notes.
GPT-3 is OpenAI’s writing algorithm. She was trained on 570 gigabytes of text. The neural network is able to write poems and news based on several words, make translations, solve anagrams and answer questions