The next highlight reel you watch could be curated by AI.
To expedite the time it takes to create multiple reels for eSports, Yahoo! researchers decide to train computers to identify key gameplay moments.
One challenge was to remove subjectivity from the task, Yale Song, senior research scientist at Yahoo! Research explains on the latest episode of our AI Podcast. To do this, researchers filtered through live stream gameplay footage, and even had Yahoo! eSports audiences participate in ranking different highlights based off of visual effects.
By developing a bot that identifies highlights in real-time, the Yahoo! team is able to deliver highlight reels almost immediately after a game ends.
“The old way of doing it … you have a team of editors watching games again and again and chopping out some interesting moments, then you pull it together. So it takes as long as the game happened,” says Song in conversation with AI Podcast host and veteran journalist Michael Copeland. “But, in this case, in our case, we don’t have to do it. There’s a machine behind looking at the scenes in real time and chopping out interesting moments.”
This deep learning system can be broadly applied, Song says. For a past research project, he applied the same technology so that AI could help identify GIF-able moments from any video link.
But, regardless of whether AI is trying to identify the best highlight reel or GIF, there is still work for researchers to do before AI can consistently curate content that is of interest to humans, says Song.
“The notion of subjectivity and trying to emulate how humans would think in novel situations, that’s something that all AI researchers are trying to solve.”
AI Podcast: AI Takes Flight
And if you missed last week’s episode, strap yourself in, because autonomous air taxis are coming. Arne Stoschek, the head of autonomous systems at A³ Airbus, shared with us the ups and downs of developing the first AI-powered, self-flying commuter vehicle.
How to Tune in to the AI Podcast
The AI Podcast is available through iTunes, DoggCatcher, Google Play Music, Overcast, PlayerFM, Podbay, Pocket Casts, PodCruncher, PodKicker, Stitcher and Soundcloud. If your favorite isn’t listed here, email us at firstname.lastname@example.org.