OpenAI Wants to Fix ChatGPT’s ‘Annoying’ Personality
If it feels like ChatGPT is laying on the flattery a little thick, you’re not the only one who’s noticed.
CEO of OpenAI, the company responsible for ChatGPT, Sam Altman said in a recent post on X the latest batch of updates made to its GPT-4o model have made it too much of a people pleaser. He added that the company is working on personality tweaks. (Disclosure: Ziff Davis, CNET’s parent company, in April filed a lawsuit against OpenAI, alleging it infringed Ziff Davis copyrights in training and operating its AI systems.)
“The last couple of GPT-4o updates have made the personality too sycophant-y and annoying (even though there are some very good parts of it), and we are working on fixes asap, some today and some this week,” he said — perhaps forgetting the word sycophantic exists.
He added that “at some point” the company will share what it’s learned from the updates. “It’s been interesting,” Altman said.
In recent weeks, OpenAI has rolled out a handful of subtle changes to the way it responds to users, such as improving its ability to guide conversations, enhancing how it listens to and follows instructions, working more collaboratively and dialing down its emoji use.
‘Losing customers is not an option’
The changes are part of a larger effort to make generative AI more intuitive and conversational in natural language, as it becomes an even bigger part of everyday life.
Reece Hayden, principal analyst at market research firm ABI Research, said companies are increasingly focused on improving the user experience to sharpen their competitive edge, so it makes sense that the personality of chatbots factors into the equation.
“An annoying experience would certainly put consumers and enterprises off usage and will need to be sorted out to ensure it remains the go-to chatbot in the market,” he said. “This remains a market which is hemorrhaging cash, and losing customers is not an option, even for a company like OpenAI with such a strong first-mover advantage.”