Late in 2022, San Francisco company Open AI released a new AI model – ChatGPT – and the world went crazy.
This version, Chat GPT 3, had been designed for use in online chatbots, and was able to conduct text conversations in a convincing conversational way.
The dialogue format meant that ChatGPT could answer follow-up questions, admit its mistakes and challenge incorrect premises.
From the point of view of writing copy, it’s relatively easy for an ordinary user to prompt it, and the quality of its output can be very good indeed.
Within a few months of Chat GPT’s API becoming available, a ton of commercial products had been developed on top of it which harness its power for specific and tightly focused marketing writing tasks.
Now we can brief Chat GPT or tools based on it, or Google’s Bard, or Microsoft’s Bing – built on the Open AI GPT4 technology – to work on blog posts, articles and other tasks requiring some desk research. While the draft it provides will almost always benefit from editing and enhancement, it can be very close to being usable.
The more experienced a user is at writing the prompts the AI requires, the better the output becomes, too.
On Open AI’s own admission, anything written using ChatGPT needs checking for truth. There is no way for the engine to determine whether the text it is compiling is correct or true!
Still, in some situations, for many users, the output from ChatGPT or tools built on its GPT4 core would be ‘good enough’ with little or no editing.