An analogy: self-driving vs. driver assist
Will robots replace editing and other jobs in the Media industry?
Artificial intelligence is perhaps the most hyped subject in technology today. So, before we get into the future of AI in the media tech space, let’s level set and consider something we’ve all thought about.
The utopian vision of self-driving cars — where I can press a button on my phone, hop in a driverless car and it takes me to the grocery store, the doctor’s appointment, or off to the airport — is still just a fantasy.
We’ve all read about the tests underway with self-driving cars and how someday humans will not need to drive, maybe even as soon as next year. During a podcast with ARK Invest on Tesla, Elon Musk claimed that:
“I think we will be ‘feature-complete’ on full self-driving this year, meaning the car will be able to find you in a parking lot, pick you up, take you all the way to your destination without an intervention this year… I am certain of that. That is not a question mark.”
But Musk’s claims should be taken in context. As more level-headed John Krafcik, CEO of Waymo, said last year, “Autonomy always will have some constraints.”
Those constraints are complex and numerous, from changing maps, construction, shadows, birds, snow and rain to the need for human interaction with other pedestrians, cyclists or cars. Not to mention the average consumers’ comfort level: how many of us want to ride in a car where we have no control?
The trials with this new technology have been marginally successful, but there are still some major hurdles before driverless vehicles are approved for mainstream use.
At the same time, driver assistance features continue to become mainstream. Many of these features have been quietly making vehicles a lot safer in the last 10-20 years.
Back up cameras, lane departure warnings, anti-lock brakes, blind spot detection, adaptive cruise control etc., these are all very effective safety features. However, in most cases these features require a human to take some action or driver assist.
AI in the media industry
Although the analogy breaks down when we consider that injury or death is not at stake in the media industry, AI has implemented technologies to make workflows more efficient. And, like cars, the creation and manipulation of media content is complex. However, AI is so complex that it will always require people.
For decades, the media industry has required technology and people to work together. AI has always been at the center of this, and it continues to have great promise. But, like the automobile industry, the risk/reward is still too great to make workflows entirely “self-driving.”
Here are a few examples of the future of AI in media tech.
Highlight Clips – driverless
At sporting events, there are numerous stationary cameras at the venue using optical character recognition. Every time the scoreboard changes, a clip could be created that goes back 20-30 seconds and the clip could then be manually trimmed to the right length for replay. This would eliminate a human looking to manually create a clip based on every score.
Content Compliance - assisted
Compliance often involves a manual process, where individuals are looking and listening, frame by frame at a scene level, to mark content in/out according to specific legal, cultural or age requirements.
Here, it’s unlikely that an autonomous AI solution could be used because the risk is too high if something is missed.
However, with an AI assist approach the suspected non-compliant content can be flagged and then further reviewed by a human. This approach limits the amount of content that requires manual viewing, saving an enormous amount of manual labor and time.
AI won’t replace people in media tech
In the end, AI can certainly assist media workflows, but it can’t replace them in many cases. The accuracy of the AI engines and risk of error is too high to just let them be autonomous.
Most of the AI engines are in the cloud and work with a specific file format. Some things to consider:
What if the content is only available on my on-premise file system?
What if my content is available in Azure, but I need to use an AI engine running in AWS?
Remember some AI engines are better than others. How do I move the content?
How do I create a video format that is only available in broadcast quality but I need a different format for a specific AI engine located in a different location?
So many situations require a human. In many scenarios, the storage is using multiple technologies (block or file) and may be in both on-premise and cloud locations.
Again, along these complex workflows, AI will be enormously helpful to take over the execution of tedious problems humans have already solved. But, as has always happened, when we humans solve one set of complex problems, the solution itself tends to create not only new possibilities but new complex problems to solve.
Inevitably, the machines will always be our assistants.