YouTube’s latest experiment uses AI to divide lengthy videos into chapters

Earlier this year, YouTube introduced a new feature called Chapters – which as the name suggests – essentially divides a video into separate chapters using timestamps. I love chapters. They save me the hassle of watching the boring parts or doing hit-and-miss adjustments on the seek bar to get to the interesting part. However, chapters can only be added by creators, so you’re essentially at their mercy to enjoy this little gift. However, that dependence might end soon, thanks to AI.

As part of its latest experiment, YouTube is banking on AI that will go through a video and identify certain visual markers to break a video into chapters. YouTube says it will rely on machine learning to recognize specific text-based signs for automatically generating video chapters. So, if a video has text cues for jumping into the next section, the AI will automatically identify it and use that time-stamp to create a chapter.

YouTube chapters are quite convenient and save a lot of time.

Say for example a smartphone review video, where the creator adds frames with prominent text such as ‘build quality’, ‘camera performance’, and more in order to jump to the next part of the review. With the new experiment, YouTube says that it wants to make it easier for people to navigate videos and quickly jump to the relevant part by taking advantage of chapters.

YouTube is currently testing the AI-generated chapters on a small set of videos. However, the company will let creators opt out of the experiment, and they can also provide some feedback on how it works or can be improved. It can be a tedious task for creators, especially those with a lot of videos on their YouTube channel, to add timestamps and chapter names to each section of their older video. This is where the AI-generated time-stamp feature comes to their aid. And of course, it is a great convenience for viewers as well.

The post YouTube’s latest experiment uses AI to divide lengthy videos into chapters appeared first on Pocketnow.