But before we go there….
Let’s understand what it is. And what it is not.
ChatGPT made its debut November 30. While the AI-powered chatbot is still in the research review phase, users can sign up and test it out free of charge.
It uses training data to give intelligent and informed responses to user’s questions.
According to the OpenAI site, ChatGPT has several capabilities:
+ It can remember what a user said earlier in the conversation
+ It allows follow-up corrections
+ It is trained to decline inappropriate requests
ChatGPT is built on the latest text-Davinci-003, a GPT-3.5 AI model made by OpenAI.
The cost per chat is “probably single-digits cents,” according to Sam Altman. This means that OpenAI is bleeding millions every day and will figure out a monetization model for it soon 😉
So what’s the catch?
The responses from ChatGPT can be outright wrong. As one user said “it will make stuff up”.
Among its listed limitations, the creators admit to the following:
– It may occasionally generate incorrect information.
– It may occasionally produce harmful instructions or biased content.
– It has limited knowledge of world and events after 2021 (when asked, it did know that in 2022, the US President is Joe Biden).
It is conceivable that over a period of time, ChatGPT trains on a larger data surface area and improves the accuracy of its models.
Until that time, ChatGPT could be a source of creativity and entertainment. Even a JayZ style song for your situation.
The ball is in your court, Google.
p.s. this article was NOT written by GPT-3 🙂