The Unprecedented Pace of AI’s Evolvement
One of the most complicated games in the world is Go, and AI first attempted the challenge of playing Go at the world championship level in 2016. The game AlphaGo was the first version that successfully beat a world-class human player. A year later, a successor version called AlphaGo Master beat the world champion at Go.
Those AI versions involved an extraordinary amount of input from previous human games for the system to learn and understand how the game Go was played. Just a year after, a further successor version called AlphaGo Zero beat AlphaGo a hundred games to zero. So, the importance of that is it took all of human history to get to 2016, where AI, for the first time, could play at a championship level. A year later, it beats the world champion, and a year after that, the version that beat the world champion is beaten by a successor version of AI 100 games to 0. And that tells you something about how fast improvements are occurring in the AI space.
The Transformative Power of AI in Media
In the area of media, AI is changing how media is produced and consumed. On the media production site, one sees AI being utilized in the creation of news. For example, Automated Insights and Toutiao in China are both using AI to be able to write news stories. A perfect example of those would either be sports results occurring in real time, stock market stories as earnings are announced, or in the case of the LA earthquake, details about the earthquake within minutes of the time the earthquake occurred, where the entirety of the news story is actually being written by AI and without any human participation.
Whether this content is embraced by people, similarly, one has, in the case of Toutiao, 120 million daily users and a recommendation engine that is providing to those users the content that is most likely to be interesting to them. And again, when one thinks about the fundamentally transformative nature of how this type of content, which was unimaginable a generation ago and not acceptable executed ten years ago, is now being delivered at scale, it creates a sense of how with accelerating change, that is likely even more fundamentally to transform the creation and delivery of news content.
Conversations on the Cutting Edge
Watch MoreWe can see the same thing in other areas of media. For example, there was a Lexus commercial that was written by IBM's Watson, and with a storyline that is somewhat amazing when one considers that it was written by AI. The script it created had a human designer of a new Lexus vehicle who watches as his car leaves the warehouse. There is then effectively a subplot where the car is going to be stolen and destroyed, and the car saves itself by using its automatic break. What is amazing about that is that one computer wrote a story that is touching to people about how another computer saves itself.
In another example, Warner Music signed a music deal for 20 albums with a German algorithm that creates music not written by people but by AI for a human audience.
Deepfakes: With Great Power Comes Great Responsibility
One other area where it is likely that AI is gonna have a fundamental and transformative effect on news and media and the consumption of media is in the area that is known as deepfakes.
Deepfakes are essentially a portmanteau word that is a combination of, on the one hand, deep learning and, on the other hand, fakes, and you put those things together — and it becomes a deepfake.
A deepfake, in practice, is essentially a synthetic combination of existing images and source images. Such an image or video, the output, will appear to be a person who is saying or doing something they actually never did. The process uses something called a generative adversarial network. And the generative adversarial network utilizes a process whose output to an audience makes it appear that a celebrity or a news anchor is saying something, but that thing was never said by that person. And the quality of those productions is getting better and better, and already only five or six years after the technology was invented, it is at a level that is close to good enough to fool an observer. The consequences over time for good and bad purposes are likely to be profound.
As an example of good use of it, the global superstar footballer David Beckham was in an anti-malaria video where he, as an English speaker, was able to state the desired message in English, but through the use of deepfakes, he was also able to provide an identical message in six languages that he does not actually speak.
That is an example of using the technology for good, but that same technology could be used in exactly the same way for malign purposes. And media companies at every level, from news organizations to news aggregators, to consumers themselves, are going to become a lot more sophisticated about what this technology is, how it is used, and its implications.
