For two evenings this week, Tuesday and Wednesday, we get a glimpse of a possible future path for broadcasting. It’s one where the machines and their Artificial Intelligence (AI) get to not only choose what we watch but also create the programme!
The BBC has been dabbling with AI to help it uncover the hidden gems in its archive. While the technology used to analyse the content both from a video and metadata perspective is not terribly new, it does keep getting better and more accurate. The smarter algorithms are now able to perform complex scene analysis and log this information as additional metadata.
This can then be searched either by humans or other algorithms to be used for a new purpose. In the case of the BBC, they have attempted to use this technology to create a program about the use of AI and computers in broadcasting, as well as to suggest other archive content for broadcast that should appeal to the BBC4 audience.
More than just recommendations
It’s a smart experiment. Most pay-TV providers have been using some form of recommendations or content discovery engine to help subscribers find the content they might like based on their viewing habits and expressed likes, but this approach will enable the BBC to trawl their vast archive and expose older content to allow it to hopefully be enjoyed again today.
Will it work? There is little reason to doubt that with enough data and the right configuration the BBC’s algorithms have ‘learned’ what BBC4 viewers like to watch and will recommend some interesting content. Viewing figures might not give an accurate picture of whether it was liked, however, as there may some additional viewers who just want to see what the experiment looks like. Hopefully, there will be some qualitative data from the BBC in the near future too.
Although the ability to offer truly personalised show suggestions is important, where this experiment will be most interesting is from a content creation perspective. We will be able to get a clearer picture of what an AI created program looks like. It will answer questions about how well can the ‘machine’ review content, create a narrative and edit the footage together to create a compelling viewing experience. The current capabilities will inevitably pave the way for AI’s future applications.
What about the future?
In future years we could expect to see an entirely computer led channel. The easiest to produce might be a new channel. By scouring the web to find breaking and trending news, a fully autonomous system could write a script, complete in-depth research (fact checked across hundreds or thousands of sources) and then a computer-generated avatar (we could even decide what avatar we wish to see and in what language) can read the news to us. It could even schedule an interviewee to join via video conference in real time, although we might question why we needed a human opinion or reaction if the computer had researched thousands of sources already!
Sports presenting could be next, as a large amount of its production is already remote, and a smart algorithm could decide which camera shot to use based on the current action on screen. Indeed with increased AR and multiple camera angles, much of this choice is being handed off to the viewer anyway. Furthermore, the commentary of some pundits and analysts amounts to little more than ‘stating the bleeding obvious’ – so not much to be missed there then!
Drama, comedy and films will see an increasing amount of AI used to draft scripts based on what was successful in the past. This is not necessarily a good thing though. We like the maverick directors and quirky writers for the programs and films we stumble across, and already these are being squeezed in a drive for instant box office or TV success.
So, all credit to the BBC for exploring this technology. Whilst you may be not a usual BBC4 viewer, this is an experiment worth observing. So, maybe for just two nights, it might be a channel worth tuning in to, and on Thursday we can reflect on whether this approach was a success or if there’s still some significant work to do.
Find out more from the BBC here.