The Chat Interface

ChatGPT has taken the world by storm. It has amassed over 100 million users in just over 2 months. Compare that to the 9 months it took TikTok to amass its first 100 million users. It’s safe to say that ChatGPT is a ground-breaking technology, the likes of which we haven’t seen before.

Part of the reason for this success is something most might not realize. The “Chat” interface is only one of several different formats that could’ve been created in order to leverage the GPT model. Indeed, if you use their online API, there’s several other formats including a pure “completion” model which just takes a single input and generates a single output. However, the chat interface is far more powerful than the completion model, and easier to use as well.

The secret sauce behind the power of this model is its back-and-forth cadence with natural language. In fact, it’s the same format that humans have used to effectively communicate since the dawn of language. We don’t need to present an idea perfectly the first time, because we can refine our thoughts and give added nuance through follow-up sentences and revisions. This is why verbal communication tends to be so effective, and also why ChatGPT feels so powerful. You can ask for “X”, and if ChatGPT doesn’t give it to you perfectly, you can add nuance or direction and watch it modify its answer based on new context.

While ChatGPT is very cool, this is only the beginning of the new era of Large Language Models (LLMs). Soon, you’ll be able to interact in natural language with all of your most-used software, enabling speed and flexibility like never before.

Need to find an image in iPhotos? Soon you’ll be able to describe the photo you’re searching for using your voice in order to find that image.

Need to move some money from Bank account A to Bank account B? Simply ask your bank app to do so.

The chat interface has stood the test of time in human communication, and now as it makes the leap to human-machine communication we’re going to see more powerful and flexible applications.