“Self-Discovery Through AI Insights”
I have been catching up from a fall break weekend in New York City (family, friends, shows, etc.) this afternoon. It’s a good thing classes don’t start for me until Thursday. While taking a break from grading and such, I stumbled across this piece on the website tech radar, “New ChatGPT prompt goes viral with Sam Altman’s approval.” It’s about a viral ChatGPT prompt:
Sam Altman, CEO of OpenAI, the makers of ChatGPT, put his name to a new viral trend on the platform recently when he retweeted it on X (formerly Twitter) with “love this” added in response to a tweet that has got ChatGPT users enthralled.
The original Tweet, from writer Tom Morgan, says simply, ‘Ask ChatGPT
“From all of our interactions what is one thing that you can tell me about myself that I may not know about myself’.
I don’t know if any of this is surprising, but it is kind of interesting, isn’t it? How long will it be before OpenAI does this?
The article goes on to describe how blown away users were by the AI response, how it made them feel better about themselves, etc. It also talks about people hacking the prompt by instead asking ChatGPT to “roast me” based on our interactions. Well, I didn’t try that, but I did try the first one. If you’re curious, here’s a link to the discussion.
I don’t want to suggest I’m the most introspective guy in the world, but I don’t thinking ChatGPT told me anything about myself that I didn’t think before. But its response was definitely flattering and clever, making connections between my cooking interests and technology, for example. So I can see why it would be a good pick-me-up for users.
Anyway, the thing I really wondered about was the title the AI gave to the discussion. ChatGPT always creates the titles of these discussions, but I thought this one, “Self-Discovery Through AI Insights” sounded a little too like marketing or advertising copy.
So I asked ChatGPT about that. It responded that no, to the best of its knowledge, the AI had not been specifically instructed to say that and that’s just now how the AI works. I pointed out that there are certain prompts about things like asking ChatGPT how to build a bomb that produce specific responses: in the case of “tell me how to make a bomb,” it explains why it cannot do that. Why couldn’t that same coding/protocols be used to require ChatGPT to title conversations from a specific prompt in a specific way? The AI said that theoretically, yes, that could happen.
Could that be used for advertising? Yes. “This would be similar to how search engines or other platforms may show ads or sponsored content. If I were programmed with specific instructions to promote a particular cruise line, I could be guided to prioritize talking about that company, its features, or its benefits when asked certain questions.”
I don’t know if any of this is surprising, but it is kind of interesting, isn’t it? How long will it be before OpenAI does this?