AI Dungeon and GPT-3: Collaborative Writing With the Help of AI

A few weeks ago, OpenAI, an AI research company based in San Francisco, California, released the third iteration of their artificial intelligence language generation tool, called GPT-3 (for Generative Pre-Training).

In a nutshell, it’s software designed to create text of anything — fiction, songs, essays, etc. You need only provide a simple prompt, and away it goes, mimicking human writing.

Want an example? On July 18, Manuel Araoz shared a post on his blog about his “experiments” with GPT-3. It’s worth a look. TLDR: It’s not perfectly convincing, but give it time. Meanwhile, if you want to give AI writing a try for yourself right now, you might want to check out the free version of AI Dungeon. It uses OpenAI’s previous release, GPT-2, and you can experience it in your browser or download it on your phone as an app.

Don’t treat it too much like a game, though. AI Dungeon is less of a Zork-style text adventure and more of a collaborative creative writing exercise with an AI that has a very short attention span.

Last month, OpenAI introduced a premium version of AI Dungeon that uses the GPT-3 model. GPT-3 is trained on 175 billion parameters, as opposed to GPT-2’s 1.5 billion, making it about 100 times more advanced. According to designer Nick Walton, GPT-3 “is just much more coherent in terms of understanding who the characters are, what they’re saying, what’s going on in the story and just being able to write an interesting and believable story.”

Not everyone’s too excited about AI language generation, though. CNBC and others have posted articles wondering if OpenAI’s research could be used “in malicious ways,” or questioning how it may affect content writing jobs in the future. OpenAI themselves have held back on fully releasing their work, as they stated in February 2019:

“Our model, called GPT-2 (a successor to GPT), was trained simply to predict the next word in 40GB of Internet text. Due to our concerns about malicious applications of the technology, we are not releasing the trained model.”

That’s the downside of AI, I guess. You take this, and deepfakes, and all the other little steps of progress we’ve been making, and you can see how strange our AI-driven future may be. And yet, it’s still pretty entertaining.


Rob Schwarz

Writer, blogger, and part-time peddler of mysterious tales. Editor-in-chief of Stranger Dimensions.

Related Articles