Even while users and researchers continue to debate how the performance of ChatGPT has changed over time, OpenAI is not letting up on adding new features to its signature generative AI chatbot product.
This week, Logan Kilpatrick, OpenAI’s first developer advocate and developer relations expert, posted on X (formerly Twitter) that a “huge set of ChatGPT updates are rolling out over the next week.”
Among the new features Kilpatrick highlighted are example prompts (already visible for the author, see screenshot below), suggested replies and follow-up questions (which Kilpatrick wrote is “very useful to reduce fatigue”), a default GPT-4 setting so that paying ChatGPT Plus subscribers don’t have to toggle on the latest/most advanced publicly available OpenAI large language model (LLM) every time they start a new chat, support for multiple file uploads when using the OpenAI Code Interpreter plugin, and more.
The initial reaction from users was mixed. While some praised the new features, including GPT-4 by default and one that allows users to “stay logged in” beyond the previous two-week period, others on the OpenAI and ChatGPT subreddit communities on Reddit asked, “Is it going to stop apologizing too? I sure hope so.”
Some even went so far as to criticize the addition of example prompts as being “terrible for someone with ADHD,” because they are distracting.
Still, it is early on in the addition of the new features, and it remains to be seen which of them users will ultimately adopt and use most in their workflows, and which fail to see uptake or are rejected in the long run.
The new features also come amid reports that OpenAI has filed a trademark application with the U.S. Patent and Trademark Office (USPTO) for GPT-5, the next LLM in its ongoing series, and that the application includes possible features such as “the artificial production of human speech and text,” and “translation text or speech from one language to another.”
