ChatGPT, the viral conversational AI chatbot developed by OpenAI, seems to be losing some of its original shine and appeal.
After ChatGPT gained tremendous popularity following its launch late last year, recent data suggests that usage and interest in ChatGPT may be declining.
Some longtime users have complained on social media and developer forums that the AI appears to be providing lower-quality answers than it was just a few weeks ago.
They describe the bot as “lazier”, “dumber” and more prone to more errors or nonsensical responses. However, OpenAI denies an intentional downgrade of ChatGPT, tweeting: “We make every new version smarter.”
The company believes that the number of users encountering more bugs is due to increased usage uncovering limitations.
From precise and expensive to fast and flawed
OpenAI’s GPT-4 AI model gained notoriety earlier this year for its remarkable capabilities, despite being slower and more expensive than other models.
It could understand images and text, making it the most powerful AI system of its time.
There is speculation that instead of one large GPT-4 model, OpenAI creates several smaller GPT-4 models focused on a specific topic.
Dubbed Mixture of Experts (MOE), this approach would reduce the computational cost of the system while providing capabilities similar to those of a single giant model.
By splitting GPT-4 into specialized smaller models that focus on narrow tasks, OpenAI could benefit from a large language model without much effort.
Response from OpenAI
After people complained about GPT-4, Peter Welinder, product manager at OpenAI, said, defended the model on twitter.
He said users may think GPT-4 is worse than before because they use it more often and see issues they hadn’t noticed before.
He insists that each new version of the GPT-4 model is smarter than the previous version.
Despite widespread speculation about the architecture of GPT-4, OpenAI hasn’t announced any major changes.
Declining user base
This perceived drop in ChatGPT’s capabilities is matched by a measurable drop in site traffic and downloads.
Data from internet analytics company Similarweb shows that global website visits fell by almost 10% in June compared to May.
iPhone app downloads have also declined since peaking in early June, according to data from Sensor Tower.
The drop in usage suggests that initial enthusiasm for AI chatbots may wane as their limitations become more apparent.
Other factors such as the completion of the academic year and concerns about regulation could play a role in the lower usage.
People worry that chatbots like ChatGPT could generate false or false information, while companies worry that sensitive business information could be leaked if entered into the chatbot.
ChatGPT’s declining user base and perceived degradation in performance suggest the viral chatbot may be struggling to maintain its initial momentum.
This trend highlights the bigger challenges chatbot creators face – as advanced as systems like ChatGPT are, they lack human context, common sense, and reasoning.
Still, ChatGPT represents a significant leap in natural language processing — even if its capabilities fell short of the potential some people expected.
Going forward, developers need to find a balance between managing expectations and pursuing the next evolution in AI chatbots.
Featured image: Vitor Miranda/Shutterstock