Streaming Media

Streaming Media on Facebook Streaming Media on Twitter Streaming Media on LinkedIn
 

Is the AI Powering Your AI-Powered Gear Really AI?

A lot of products tagged as "AI-powered" were doing exactly the same thing before AI. But labeling their familiar features as "powered-by-AI" makes the same tech seem fresh and innovative.

It's that time of year again. As I write this, BirdDog is teasing "X", and Blackmagic Design has scheduled a livestream announcement just before NAB, where everyone will get to see what new gear they bring to market that surprises everyone. I'm getting releases from Canon, Sony, Ross, Haivision, Rhode & Schwarz, Western Digital, etc. The list goes on.

Streaming-capable gear is not something special anymore. It's common. Now all the buzz is around AI features. AI-this, AI-that, up to and including AI CEOs. It really feels like when the industry moved from standard-definition to HDTV. Everything was HD-this, HD-that. I did a column called "HD Today" for six years. Even household appliances were HD because they were, better, somehow. Now we have AI-enabled clothes washing machines and AI microwaves, so called ecause these devices can "sense" what to do.

But wait, weren't they doing that before AI?

It used to be called "if/then" programming. Like the basic thermostat in your house--if the temperature gives above 75, then turn on the AC. There's also If This, Then That (IFTTT) devices that anyone can use to automate their home, their systems, etc. Machine learning is sort of an automated IFTTT, like the Nest home thermostat that learns the household routine and if nobody is home, then it doesn't need to cool the house as much.

A lot of products touting "AI-powered" were doing exactly the same thing before AI. Motion tracking in PTZ cameras predates AI by over a decade. Even motion tracking that can follow a person, behind other objects, even if they turn away from the camera were doing this long before "AI" made it possible. But today, we have AI tracking that does the same thing.

Video mixers that could auto switch, even follow the person speaking automatically--also predates AI by well over a decade. We used to call it Video Follow Audio: cut to video associated with the microphone that has the loudest audio. But today, it's an "AI" feature. Because this labeling makes the same tech seem fresh and innovative.

How about the high-resolution camera with "AI" technologies that automatically track and dynamically resize the frame to include the person? Apple calls this their Center Stage feature. Is this AI? No. It's pre-programming: If there's a face or two, if there's movement around or connected to the face, then resize the active crop to fill the frame with the subject(s). IF, then.

AI is really located where the answers can't be preset. Can't be pre-programmed. Where it is simply not known. Then it takes Intelligence to create a new answer, action, solution that was not intended. Whether computer or human generated, it's the unforeseen answer, the result that pushes in a new generation. That's where AI really comes into play, and this truly does exist. But AI is way more limited than the many solutions claiming "AI."

NAB is one of the biggest broadcaster & streaming shows in the world, and it kicks off this weekend. We'll see and hear a lot about all sorts of products "integrating AI" to make things easier for producers. If you step back and look at the market, however, you'll see much of these touted solutions were around before the buzzword took hold, and the solutions themselves can often be reduced to if-then programming. If the signal is not there, then put up this graphic, is not AI.

Using computer tracking to follow a soccer ball that then mistakes a bald referee's head for the ball, is not innovation; it was trying to save the cost of a real camera operator who would not have made that mistake. It was not intelligence at al. It was if/then programming gone wrong.

There will certainly be a lot of innovation, at NAB and beyond. Just be sure to assess tools and solutions on what they can do for you, not on whether they have the letters "A" and "I" associated with it. The key is to determine where the innovation truly delivers value, increased usability, reduction of workload, or improved deliverables. Then use your organic intelligence to make a decision based on those answers.

Related Articles
NDI runs the risk of fumbling the ball on the 20-yard line—squandering the nearly decade lead they have in IP video by not enforcing that licensees integrate all of the standard. I write this in the hope that they re-establish what the NDI standard means: that to license NDI, and display the NDI badge, a product must be 100% compatible.
If you're looking for a compact PTZ camera system for a small performance stage, conference room, podcast table, demonstration kitchen, instruction classroom, or any number of setups where the cameras can be positioned close to the subjects, and you would really benefit from the ability to manually change the shot or leverage automatic tracking, the OBSBOT Tail Air is custom-tailored to your needs.
Mark Alamares of Mobeon, one of the streaming and virtual production industry's most innovative media companies, is stepping aside as CEO and handing the company's operational and strategic planning reins to Chingiss Tron, a virtual executive who may be the streaming industry's first AI CEO. In this interview with Streaming Media editor Steve Nathans-Kelly, Mark (who will stay on Mobeon's COO) explains what having an AI CEO is all about, and why he believes other media companies will see the strategic and operational benefits and follow Mobeon down this path in the years to come.