Comply with ZDNET: Add us as a preferred source on Google.
ZDNET’s key takeaways
- OpenAI has up to date ChatGPT’s personalization web page.
- The web page contains character choices, reminiscence toggles, and extra.
- Not all customers really feel seen or heard by the replace.
OpenAI is making it simpler for ChatGPT customers to customise what it calls the chatbot’s “character” and communication type.
ChatGPT now presents an up to date personalization web page, which OpenAI CEO Sam Altman previewed in a Tuesday X post. The brand new web page features a dropdown menu with a spread of character varieties, together with “Cynic,” “Robotic, “Listener,” and “Nerd,” and a “customized directions” immediate part, by means of which customers can modify the chatbot’s outputs with requests like “Keep away from millennial jargon.”
Additionally: AI models know when they’re being tested – and change their behavior, research shows
There are additionally sections the place you’ll be able to specify your most well-liked nickname and occupation, and one other the place you’ll be able to record private “pursuits, values, or preferences” — the aim of which seems to be to create a conversational expertise that feels extra like chatting with a trusted buddy or colleague than with an unfeeling machine. Towards the underside of the up to date web page, which is accessible now by way of settings in ChatGPT, you’ll be able to modify the chatbot’s reminiscence capabilities.
The science (and enterprise) of synthetic personalities
The refinement of chatbot personalities has develop into a essential focus for tech builders as they proceed to promote AI to people and companies. The problem is to construct programs that generate correct info in a method that additionally feels participating and personable. This has often gone too far, as within the latest case of ChatGPT customers complaining that the chatbot’s responses tended to veer into sycophancy and extreme flattery.
Additionally: ChatGPT will verify your age soon, in attempt to protect teen users
Completely different corporations have additionally taken varied approaches to synthetic personalities: xAI’s Grok tends to be irreverent and edgy, Claude is extra of a sober and cautious digital assistant, and ChatGPT’s tenor tends to be extra versatile relying on the preferences of every consumer.
Specializing in personalization
That flexibility and adaptableness have develop into an even bigger focus for OpenAI within the aftermath of final month’s debut of GPT-5, the newest giant language mannequin (LLM) to energy ChatGPT. Many customers have complained for the reason that mannequin’s launch that it is inferior to its predecessor, GPT-4o, each by way of its pace and its communication type, which some discovered to be flat and chilly.
OpenAI shortly responded to these complaints with a flurry of updates to GPT-5, one among which allowed paid ChatGPT Plus subscribers to regain easy accessibility to 4o.
Additionally: Is ChatGPT Plus still worth $20 when the free version offers so much – including GPT-5?
Altman stated that one of many central classes of GPT-5’s underwhelming preliminary success was that OpenAI wanted to focus extra on constructing extra personalised and customizable instruments.
“Lengthy-term, this has bolstered that we actually want good methods for various customers to customise issues (we perceive that there is not one mannequin that works for everybody, and we have now been investing in steerability analysis and launched a analysis preview of various personalities),” he wrote in a X post on Aug. 8, including the instance of some customers maybe preferring plenty of emoji whereas others “by no means need to see one.”
OpenAI obtained an identical backlash from many amongst its userbase when it introduced earlier this month that it could retire ChatGPT’s Standard Voice Mode. The corporate later rescinded that call.
Additionally: How people actually use ChatGPT vs Claude – and what the differences tell us
(Disclosure: Ziff Davis, ZDNET’s mother or father firm, filed an April 2025 lawsuit in opposition to OpenAI, alleging it infringed Ziff Davis copyrights in coaching and working its AI programs.)
Extra calls to reinstate 4o
ChatGPT’s new personalization web page has already obtained a cold reception on-line, with many individuals arguing that the corporate is misunderstanding their complaints about GPT-5.
“A clear consumer interface is welcome, thanks, however please additionally think about the request from 1000’s of customers to maintain 4o and Commonplace Voice Mode long run,” one person wrote beneath Altman’s X publish. “4o is ready to intuitively adapt to consumer intention and choice, dynamically working alongside you. It does not want customized directions or personalisation.”
Additionally: I did 24 days of coding in 12 hours with a $20 AI tool – but there’s one big pitfall
One other particular person’s reply put it much more succinctly: “No mixture of toggles and directions can replicate the natural and easy vibe of interacting with 4o.”
OpenAI has no plans to reinstate 4o for customers of ChatGPT’s free tier, an organization spokesperson confirmed to ZDNET.
