\n
![](\"https:\/\/cdn.arstechnica.net\/wp-content\/uploads\/2025\/09\/branch_2-1152x648.jpg\")
<\/p>\n
\n<\/p>\n
On Thursday, OpenAI introduced<\/a> that ChatGPT customers can now department conversations into a number of parallel threads, serving as a helpful reminder that AI chatbots aren’t folks with fastened viewpoints<\/a> however quite malleable instruments you may rewind and redirect. The corporate launched<\/a> the function for all logged-in internet customers following years of person requests for the aptitude.<\/p>\n The function works by letting customers hover over any message in a ChatGPT dialog, click on “Extra actions,” and choose “Department in new chat.” This creates a brand new dialog thread that features all of the dialog historical past as much as that particular level, whereas preserving the unique dialog intact.<\/p>\n Consider it nearly like creating a brand new copy of a “doc” to edit whereas protecting the unique model protected\u2014besides that “doc” is an ongoing AI dialog with all its amassed context. For instance, a advertising group brainstorming advert copy can now create separate branches to check a proper tone, a humorous strategy, or a wholly completely different technique\u2014all stemming from the identical preliminary setup.<\/p>\n