If you’ve been following the tech news over the last few weeks, you’ve probably seen several mentions of something called Chat Generative Pre-trained Transformer, or ChatGPT. To debunk any confusion about what it is, I put together this introduction to help you out.
I just used ChatGPT to:
1. build a list of US cities
2. convert the list to JSON
3. output a CSV file pic.twitter.com/Y2sR86ampw
— NoCodeDevs (@nocodedevs) December 24, 2022
Generative pre-trained transformer
ChatGPT is owned by OpenAI and was created using the RLHF method. OpenAI developed the platform to generate human-like responses in a conversational context. For example, if you ask it a question on any subject, like marketing, it’ll give you a comprehensive answer. The answers are derived from the terabytes of data out there on the internet.
The software gained instant popularity and a million users after just one day. Even given the rising popularity of chatbots among both enterprises and consumers, that’s a very impressive figure. Here’s how ChatGPT works in practice.
To test the ChatGPT prototype, I asked two questions in the chat. The first one was “What are your predictions for content marketing for 2023?”
Here is the response I received:
This is a very comprehensive answer and the kind I would usually expect to receive from marketing experts. However, you see that ChatGPT does a disclaimer at the start saying that it doesn’t have the ability to make predictions for the future.
Indeed, NLP models aren’t currently capable of that, as they thrive on ML and data, not abstract possibilities. Even OpenAI admit in their blog that ChatGPT has limitations, such as being excessively verbose and overusing specific phrases. For that reason, you should be careful when using the software – not least because it’s still a prototype.
Speaking of abstract possibilities, I also asked ChatGPT a more obscure question – “What is the meaning of life?” And this is the answer I got:
Overall, the response is pretty much what I expected – a well-structured answer that effectively provides an overview of some well-known existing opinions on the subject. If improved further, ChatGPT would have a lot to offer businesses.
Currently, a lot of people are using ChatGPT for software support, including NoCode. SaaS support requests and streamlining HR processes and lead generation are only a few of many emerging and potential use cases. Some thought leaders even believe that ChatGPT will make copywriters disappear. As a writer, I’m hesitant to go that far – at least at this stage.
Photo credit: The feature image has been taken by Iryna Khabliuk. The images in the body of the article have been taken by the author for TechAcute.
Sources: OpenAI blog / Stephen Shankland (CNET) / Bernard Marr (Forbes)