OpenAI’s hugely popular AI chatbot, ChatGPT, recently received an impressive upgrade with the release of the O1 preview – a new line of models that de-emphasize responsiveness and instead focus on more “thoughtful” and considered solutions to user input.
While the company’s latest preview represents a groundbreaking attempt to improve performance and accuracy in answering complex scientific, coding and mathematical problems, ChatGPT may be working on another feature that could completely change the dynamics of AI chatbots.
According to some users, OpenAI’s chatbot has started to socialize and initiate conversations without requested.
ChatGPT masters the icebreaker
Did you just message me first?
ChatGPT user, SentuBill
In a post shared on RedditUser SentuBill (a free ChatGPT user who signed up for the beta program) reported that he opened the app as usual yesterday and noticed that a new conversation was in progress. The problem? The conversation was not initiated by SentuBill, but by ChatGPT.
The opening message of the conversation, which SentuBill has since shared to prove its authenticity, asks the user, “How was your first week of high school? Did you settle in well?”
While this would be a completely harmless question by any other standards, the fact that ChatGPT was able to turn the tables and prompt the user to talk was quite surprising to the Redditor. “Did you just message me first?” SentuBill replied, before OpenAI’s chatbot confirmed, “Yes, I did!”
A more talkative GPT
According to comments in the original thread, SentuBill is not the only user experiencing this change in behavior.
While several users in SentuBill’s thread reported similar interactions with ChatGPT over the past week, one X user shared a video of ChatGPT starting the conversation immediately after enabling enhanced voice mode.
Yo I can’t believe I got this record, it started the conversation on its own, Jesus🤯what the hell??? pic.twitter.com/u8QrPsbODi16 September 2024
Is it a bug or a feature?
Aside from the possibility that the chat was manipulated because SentuBill shared the conversation, is ChatGPT’s sudden mastery of the icebreaker a bug or a new feature? LaptopMag has reached out to OpenAI for comment and further details on this matter.
However, given the complexity of the message created (which requires the chatbot to use its save feature and start a new conversation), it is very likely that these users were recruited for some sort of A/B testing of a new feature that may be included in the chatbot in the near future.
In A/B testing, a company uses two versions of the same product or service that differ in a specific way to see which one is best received by customers or users.
In this case, a select number of ChatGPT users may find that their chatbot model has the additional functionality of being able to actively seek out conversations rather than just responding to user prompts.
OpenAI would likely use such a feature to encourage interaction with its chatbot by giving users a little nudge or reminder that ChatGPT is ready and waiting to help, but it could also herald some broader expansions to the chatbot’s capabilities.
What does this mean?
Traditionally, there are large language models like ChatGPT that provide natural language responses to user input, be it through text or audio interactions. However, this new method of chatbot-oriented communication is something completely different.
While the O1 preview may be making headlines for setting an entirely new performance record for OpenAI’s chatbot, for the average user, ChatGPT’s simple ability to start a conversation could have an even bigger impact.
Like ChatGPT’s enhanced voice mode, this new feature would help make conversations with the chatbot feel more like an artificial companion. This could have a positive impact for those who use OpenAI’s chatbot to alleviate loneliness symptoms or as a memory aid.
This new capability could also unlock a number of new features, with ChatGPT then being able to remind users of specific appointments or unfinished work, and even participate in scheduled tasks and communicate the results to users.
What are the risks?
On the other hand, such a move could have negative implications. Before introducing ChatGPT’s enhanced voice mode, OpenAI expressed concern in a System Card report published in August that a more human voice could cause users to become emotionally dependent on its chatbot.
During testing, OpenAI observed that certain users formed “common bonds” with its AI, leading the company to admit that “further research is needed to understand how these effects might manifest over longer periods of time.” However, as enhanced voice mode continues to roll out globally, this new feature could lead to similar issues.
A feature like this, which allows the chatbot to actively seek a user’s engagement and attention, could humanize the company’s software even more in the eyes of many, making those emotional connections feel like mutual connections.
ChatGPT’s latest feature in testing also suggests that the chatbot has agency and freedom to communicate in a more “lifelike” way. This could indicate that it has a desire to speak to a user specifically, rather than being pre-programmed to respond to every user.
If OpenAI was concerned about the impact of its enhanced voice mode, then there is undoubtedly reason for the company to continue to closely monitor interactions during these current tests.
outlook
While commenters on the original thread are quick to point out how creepy this new ChatGPT feature is, they are no less impressed, with many users admitting to having longed for such a feature in the chatbot for some time.
Whether it will become a permanent feature in future versions of the chatbot remains to be seen, and OpenAI will no doubt be closely monitoring how select users respond to ChatGPT’s richer messaging features before making any final decisions.
However, it is another example of the company’s forward-thinking approach to artificial intelligence and human-computer interactions, and opens the door for other models that can once again follow in ChatGPT’s footsteps.
More from Laptop Mag
