I won't go into all the details here ... but I asked ChatGPT to pretend to be a first-year Forrester Research analyst and offer three ideas for saving Bed Bath & Beyond from Bankruptcy. None of the solutions included merchandise, the product BB&B actually sold.
I then asked ChatGPT why it didn't offer any merchandise solutions? It responded by offering me merchandise solutions. I asked the app if it didn't offer me merchandising solutions because it was pretending to be a first-year Forrester Research analyst? The app told me that "it did originally offer me merchandise solutions."
I told ChatGPT that the statement was a lie - it originally offered me marketing solutions.
ChatGPT then said "you are correct, I only originally offered you marketing solutions."
In other words, I was able to catch ChatGPT in a lie, a lie the application admitted to.
What happens when humans are not smart enough to catch AI in a lie?
Do not cede your corner of the world to AI, and if you choose to do just that, do so at your own peril. Don't doom us.
P.S.: Log into your account and give it a try: https://chat.openai.com/auth/login
No comments:
Post a Comment
Note: Only a member of this blog may post a comment.