By Rex Wong, Founder and CEO of AIVON
Two years ago, a strange little online movie called Sunspring caught the attention of a number of media outlets. The nine-minute video, in which characters say grammatically and conceptually questionable lines like, “I just wanted to tell you that I was much better than he did,” appeared in stories everywhere, from TechCrunch to Slate to CBS News. What made this little oddity so interesting? It claimed to be the first movie written by an artificial intelligence, which had been fed scripts from the likes of such sci-fi classics as Ghostbusters and The Fifth Element.
While an intriguing concept, the film itself, even at only nine minutes, is, well, pretty unwatchable. That’s really no surprise. AI’s greatest strength isn’t necessarily in replicating what we think of as “creative genius” – even our best and most optimistic AI scientists are still in the very earliest stages of trying to reproduce the “secret sauce” in our greatest artists’ brains that makes such creation possible – but rather in parsing and moving through an established decision or answer tree much more quickly than any single person could.
This is why IBM’s Watson computer so easily crushed its human competition on the game show Jeopardy and why the accident rate for self-driving cars, even in their infancy and even without the advantage of evolved human “instinct” behind the wheel, is already so much better than current human-driven (and human error-driven) rates.
While it’s perhaps noble and even endearing (or depending on your perspective, scary) to forge an artificial intelligence “in our own image” – that can “feel” creative inspiration or happiness or sadness or anger or love or any of the sensations we associate with being “human” – this perhaps ignores what AI has best done so far and probably what it will continue doing best.
It is easy to become enraptured by our best speculative science fiction about the nature of a machine’s soul in human-scripted movies like “Blade Runner,” “Ex Machina,” or “Her,” that we fail to see what AI’s true purpose for society, at least in the near future of our lifetimes, which boils down to this: quickly doing those tasks humans cannot easily or willingly do, or where human error regularly presents disastrous and even fatal consequences (e.g. self-driving cars).
Metadata is the medium
So where specifically will this impact first be immediately felt? In the online world at least, where that most coveted of demos (the young consumer 13 to 24 years old) watches 2.5 times more content than on traditional television, the answer is metadata. As the metadata associated with online video becomes fuller, richer and more complex, the requirements needed to “surf” that metadata and associated content will also become more technically complex, even if we must keep the frame of such searching simple and easy for the user.
I was the founder of Applied Semantics, which later become AdSense, the company subsequently acquired by Google and which would form the backbone of its advertising business, which most estimates peg at being about three-fourths of Google’s total business. While the ad business was key to keeping Google flush with revenue and profit, we can’t underestimate Google’s initial value proposition itself: it was perfectly pitched to take advantage of the internet’s first wave of personalization, text-based search.
Coupled with hyperlinking, the ability to crawl and search the web is what made our personal computers, and eventually our smartphones, true knowledge engines. Google’s effectiveness led it to become synonymous with the internet itself, as the U.S. president or anyone who has recently “Googled” themselves could tell you.
This kind of “great leap forward” for non-textual content, like video, still awaits. On YouTube alone, users upload 400 hours of video each minute and, other than for the broadest and most simple searches, there is just no good way to catalogue it all. There is a raft of problems with that first basic part of “crawling” for video search, so dependent currently on really shoddy tagging, or even a complete inattention to such tagging. Users come up with their own tags that may or may not be apt and tags are hijacked or otherwise misused, such as inappropriate or explicit videos that are tagged with “kid-friendly” subjects.
Concurrently, the YouTube revolution has given rise to a number of different online video hosting platforms, and this fragmentation and decentralization will only accelerate as the producers of the content look to ensure their cut of the pie. Producers of popular entertainment, especially large media conglomerates like Disney, will begin trying to “out-Netflix” the Netflixes of the world, making sure not to make the same mistake record labels made with Apple and iTunes at the beginning of this century as they handed over the keys for digital distribution because they were too lazy or short-sighted to figure it out for themselves.
Such fragmentation and cross-platform siloing will only confuse search even further. However, metadata will remain as the shared language among these platforms and this is where third-party providers, combining the insurgent technologies of AI and the blockchain, might find their way in to the massively disruptive and lucrative online video revolution to come.
So where will AI fit into this continually changing ecosystem? With the right nudge from us, it could change the way we search for and consume what we think of as “video content.”
Who made the most money from entertainment in the previous century? Even more than those who actually created the entertainment, it was the producers: those movie studios, those television production companies and broadcasters, those record labels, etc. that shepherded all this content. They were the capitalists who could marshal the resources needed to make the business model of entertainment work, so crucial in the last century, but which is now shifting or outright dying in our connected, decentralized world.
Yes, the platforms hosting this content will still make their money through subscriptions, ad revenue, or a combination of the two, even in this decentralized world, but with so much content, there will also be value placed on cataloguing and curation. Some of that could come from the “human cloud” of curators, some of that could come from AI, and some of it could come from a combination of both.
On the “people side,” this will require massive curation muscle power. Driving this effort will be the power of crypto and the blockchain to incentivize what I call “the human cloud.” We’ve already had something like this with amateur expert editors on Wikipedia and amateur creators on YouTube, but crypto will make it even easier for such expert efforts to be rewarded on a micro-transactional level. Such incentives will be all the more enticing as traditional manufacturing and other jobs are lost to automation and AI in the enterprise and greater value is placed on specific, expert knowledge. Better “human-powered” video curation and indexing will better “teach” the AI’s own data decision tree, eventually making video as easily crawlable and searchable as text on the web.
With apologies to Sunspring’s AI screenwriter, I’d be very surprised if artificial intelligence writes, directs or stars in the perfect film, or even a pretty good one, in our lifetime. But, with the right amount of coordination alongside focused human curation, it might just help us easily index and find that perfect film.
About the author:
Rex Wong is a serial entrepreneur with over 25 years of experience in the Internet, technology, media and software industries. Rex is currently the CEO & Co-Founder of AIVON, the first blockchain protocol and ecosystem for building decentralized video applications leveraging AI metadata, which is building the first Open Video Search Engine. Rex is also Chairman & Co-Founder of iVideoSmart, a global Internet TV network with over 500 million addressable users.
Previously, Rex was the CEO of uCast (previously known as Qello), an OTT Video Platform technology provider, CEO & Co-Founder of 8sian Media (acquired 2017), the leading Asian multi-channel network with over 5 billion video views in 2017, Co-Founder & CEO of DAVE.TV (acquired 2008), one of the first IPTV platforms; and founding investor of Applied Semantics (acquired by Google, 2003), which developed AdSense. Rex has founded and run multiple startups, achieving numerous successful exits thus creating shareholder value in excess of a billion dollars. Rex’s companies have won numerous awards including Best of CES, Best of Internet World, Red Herring 100 and Always On 100.