GPT-3 is a neural network machine learning model trained using internet data to generate any type of text. Developed by OpenAI, it applies machine learning to generate various types of content, including stories, code, legal documents, and even translations based on just a few input words. GPT-3 has been getting a lot of attention for the seemingly unlimited range of possibilities it offers. GPT-3 is also being used for automated conversational tasks, responding to any text. So here mentioned the 10 experiments with GPT-3.
Interviewing AI: Using the Chat preset within GPT-3 Playground you can ask the current entity about its personality. And while of your dialog, the GPT-3's personality emerges. Note that after 2048 tokens there's a hard cut, and you never will encounter the same personality setting again. It imitates a human person worrying about data privacy.
Doctor's Assistant: The AI has been fed with patient files, describing their profile and symptoms in a few lines. The AI spontaneously makes suggestions of what the disease could be. GPT-3 got away with an impressive 8 out of 10 correct guesses. This could become amazing support to doctors, and a great tool to investigate.
Writing Loveletters: This is an obvious exercise from Creative Writing: imagine being somebody and write a love letter from this entity. GPT-3 is not an AGI, but it's capable of comprehension. It can understand, what is a toaster, what is a love letter and which toaster features could be reflected in a love letter written by the latter.
Dungeons and dragons: Researchers' nerdy side could not resist a GPT-3 based game of dungeons and dragons. The AI keeps storytelling and asking you about your next move. Unfortunately, researchers didn't make the right choices and it got very dark very quickly.
Style Imitation as letter exchange: In another experiment, let famous persons write letters to each other. Here Jane Austen congratulated Goethe on their Birthday, but it escalated quickly.
Dealing with Management: Many tried to use GPT-3 to rephrase sentences using different tones and styles but we got someone who railed down an approach that could improve many lives: the toxic manager translator.
Write short stories or fairy tales: Self-attention-driven transformers keep the stylistic coherence, and so can let GPT-3 generate short stories, piece after piece. In this experiment, the researcher tried Short Dreams before you wake up in various languages and was very satisfied with the result.
Email Literature: An experiment aims to be a ready-for-market product in no time: the email writer. Other side AI wants to perfect, using GPT-3, an AI who will be capable of learning writing style and turning any bullet-point list feed into a full written email.
Find non-obvious connections: This one is great for brainstorming finding two random topics without direct correlation and letting GPT-3 discover connections, like between McDonald's and Dostoyevsky's Crime and Punishment.
Beyond translation: GPT-3 has no issue understanding how to turn a non-direct message into an intelligible translation for people with high functioning autism. The capacity GPT-3 has to frame the finesse between an implied message and a direct one impressed us.