Roughly 20 hours ago, OpenAI published a new conversational language model called ChatGPT. According to OpenAI, the dialogue format makes it possible for ChatGPT to answer followup questions, admit mistakes, challenge incorrect premises, and reject inappropriate requests…
…so naturally I decided to have a go at it.
ChatGPT is pretty smart and conversational
It seems like ChatGPT is pretty knowledgeable and is able to explain stuff quite elegantly.
Not only that, as you can see, ChatGPT is also pretty good at not engaging in sensitive topics, or any speculative matters.
In the example above, it explains to me it is programmed to not produce any responses that involve speculation. Even if I prompt it, multiple times, to just give it a go, it won’t.
It also recognizes when I’m being a little bit more deceptive:
So what is the purpose of it?
My first impression is that ChatGPT steers away very consistently from sensitive topics, giving advice, and any forms of speculation. So, what can you do with it, other than ask it for interesting facts about the world in a conversational way?
Well, I asked it to write me a synopsis for a story and ChatGPT came up with the following:
Not bad. Not bad, at all. For lack of a better word, I would even describe it as poetic.
Then I asked for the first paragraph. And a second one.
And, finally, to finish the story for me.
Some observations
There’s a few things I’d like to point out:
It successfully came up with a story with multiple characters with a beginning, middle and end
The story was consistent with details from the prior given synopsis
The story included dialogue between characters
Apparently butterflies can look back and smile
All jokes aside, the level of detail, the consistency, and the attention span that is held throughout the story is remarkable.
…
That’s all I have for you (as of right now). I highly recommend checking it out for yourself.
It’s free (for now). Visit chat.openai.com