Unsurprisingly, the AWS booth at NAB was heavy on AI. Amazon needs to sell a lot of it.
One of the featured showcases on the booth was for AWS Elemental Inference. I will say that in my experience, AI is pretty crap at inference. But that’s not the kind of inference AWS means. In their case, it’s a new fully managed service that applies AI in parallel with live video encoding, so broadcasters and streamers can automate 16:9 vertical video creation in a reported 6-10 seconds. It maybe “infers” the aspect ratio? With Gen-Z viewers apparently consuming 88% of streaming content on smartphones, this shift to mobile-first, AI-powered distribution would be a good thing to automate. So far, this has been the kind of thing existing AWS customers like Fox and NBCUniversal use alongside their existing AWS media services. So probably not for the little people. However, this example of AI automating processes like mobile conversion is the kind of task AI can do well.
The Cloud Court Challenge activation in the West Hall Lobby showed a more entertainment-oriented application of AI, the kind of thing you might see in a theme park or museum. Attendees put their basketball skills to the test, and got cloud and AI influence on their performance, receiving real-time performance insights through detailed data analysis and computer vision, plus their own player card and social video to take home. The emphasis on the sports fan experience was very on the nose at Amazon’s Sports Theater too, where they sold it hard, both as a tool for doing sports production and as an engine for new fan experiences. I think one of the more fascinating applications came from the PGA Tour, where real-time shot data and player performance metrics trigger automated production decisions.
NAB promoted that more than 1,000 exhibitors brought AI-powered production and cloud-based workflows to the show, and dropped names including Adobe, Blackmagic Design, Canon, Dell, Intel, Microsoft, NVIDIA, and Sony.
PTZ Optics debuted its Visual Reasoning initiative. This includes AI-driven auto-framing and multi-target tracking on their Move 4K cameras, which can analyze live video to trigger automated actions without a dedicated camera operator. Panasonic Connect also leaned heavily into AI for PTZ. For-A showed their MixBoard, a software-based switcher designed for corporate events and live presentations. It’s described as using AI to automatically generate and draw system diagrams, lower thirds, titles, and real-time graphics for multi-layer presentations. Audinate demonstrated how AI can assist network optimization and device discovery, though that, for me, fell under the banner that most AV processing can be called AI in some fashion.
So far, so meh on AI. But it is not going away. Sometimes it will just be another word for intelligent signal processing, or software programs; sometimes it will help with creative brainstorming, or automate tedious tasks well enough that, with oversight, it can be useful. We will be learning together if it can ever save us any time, or if it’s really just entertainment for when you want to noodle about something you don’t need accurate answers to. I do think everyone should use it to really understand what AI can do well (some things) and where it has a very long way to go, and no real incentive to get anywhere useful. It will be a drift engine in an amount of applications, and probably an agent of some frightening mistakes. But it can’t really be ignored, so I’m going for “educated skeptic”.
