The OpenAI API can be used to help with lots of different tasks that involve understanding or creating language or code. There are different existing applications available, and you can also make your own custom applications. These can be used for things like generating new content or organizing information into groups. It uses the same AI as our friend DALL-E (GTP-3), just to process words rather than images.
So far, so uninteresting.
However, the AI which runs the OpenAI API is incredibly smart. And unbelievable simple to programme. Given any text prompt, GPT-3 writes copy in natural language. In fact, it has written most of this post so far.
Here’s an example of a recipe it created:
Write a recipe based on these ingredients: Rice Krispies Chicken Eggs Flour Milk Onions Chillies Instructions: 1. Preheat the oven to 200 degrees Celsius. 2. In a bowl, mix together the rice krispies, chicken, eggs, flour, milk, onions and chillies. 3. Using your hands, form the mixture into small nuggets. 4. Place the nuggets on a baking tray lined with baking paper and bake for 15-20 minutes or until golden brown. 5. Serve with your favourite dipping sauce.
Why should you care?
We all know that we’re going to be replaced by robots one day (at least I certainly do). But until such a time as general purpose AI exists, then open source tools like this are the closest thing we have to a sentient hive mind. The OpenAI API can ‘think’ at a speed and scale beyond any one person.
A large community of creators have asked it to do everything from processing massive sets of content, to turning movie names into emoji’s.
Most importantly, it is only as useful, creative or entertaining as the question it is asked. Which could be anything. Much the same as all our work, the brief is critical.
It’s surprisingly easy to use. You can give it a go here.