OpenAI says ChatGPT messaging first was a bug, not a new feature

[Pelican Press 0]

This is the hidden content, please

• Sign In
• or
• Sign Up

OpenAI says ChatGPT messaging first was a bug, not a new feature

ChatGPT’s latest model o1 – or Strawberry as it was codenamed – has begun rolling out and it promises to make the chatbot AI more human-like than ever. That includes the ability to better understand natural language, but few people expected ChatGPT to be able to message you first. For a brief time it seemingly could, though that was apparently a bug rather than a hint at a new upgrade.

Several users reported on social media that ChatGPT had started a conversation with them; in

This is the hidden content, please

• Sign In
• or
• Sign Up

ChatGPT asked the user how their first week at high school went; when it was asked if it messaged first, the Open AI **** proudly responded “yes”, later expanding that it can now initiate a conversation to follow up on things it has previously discussed. Another user claimed in the comments that the AI asked them about health symptoms they had previously discussed with it.

As proof of the new feature the user also

This is the hidden content, please

• Sign In
• or
• Sign Up

, to show that ChatGPT did indeed initiate the conversation unprompted. Though some were skeptical, with one

This is the hidden content, please

• Sign In
• or
• Sign Up

showing how the chat could have been faked.

Regardless, people were excited for this kind of feature, real or not. Checking in after your first day of high school or on your health demonstrates a (somewhat creepy) sense of human-like compassion, but we’re thinking of the more mundane reminders.

How often have you researched something like plans for an upcoming trip then forgotten about everything you looked up – ChatGPT could now remind you to book a table at that restaurant it recommended or plan your visit to that famous landmark you must see so your planning efforts don’t go to waste.

However, it appears that this was a bug, not an intended addition to ChatGPT that’s being A/B tested.

Gone but not forgotten

data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///ywAAAAAAQABAAACAUwAOw==

(Image credit: Future)

OpenAI has now responded, telling

This is the hidden content, please

• Sign In
• or
• Sign Up

in a statement that it had “addressed an issue where it appeared as though ChatGPT was starting new conversations,” adding that “this issue occurred when the model was trying to respond to a message that didn’t send properly and appeared blank. As a result, it either gave a generic response or drew on ChatGPT’s memory.”

Sign up for breaking news, reviews, opinion, top tech deals, and more.

That means, at least for now, ChatGPT won’t be able to initiate a conversation. Though given the level of buzz this situation has created we wouldn’t be shocked to see OpenAI start to work on something similar.

If the **** already has the ability to recall events from previous conversations there’s no reason why some extra logic programming couldn’t get it to start chats based on what it knows – provided you give it permission to. For now, we’ll have to wait and see where OpenAI takes ChatGPT next.

You might also like

This is the hidden content, please

• Sign In
• or
• Sign Up

#OpenAI #ChatGPT #messaging #bug #feature

This is the hidden content, please

• Sign In
• or
• Sign Up

This is the hidden content, please

• Sign In
• or
• Sign Up