OpenAI, the AI-research organization that has mostly been famous for being tangentially related to Elon Musk had published an AI model called GPT-2, that was successfully creating fake news articles with a rate of XX%. The model has also been re-trained to create tweets, stories and novels, etc. by enthusiasts and professionals all over the world.
The same OpenAI, in late April, published Jukebox, “a neural net that generates music, including rudimentary singing, as raw audio in a variety of genres and artist styles.” Listening to the samples produced makes you almost believe in magic.
If we are able to create neural nets or other AIs that can create stories, art, video, music, we might also be able to create an AI that makes movies – even possibly starring long-dead actors. Imagine a movie with Bruce Lee set in space. All created without the intervention of people, by electric signals in a server farm somewhere.
If the movie and music industry falls to AI (much as the long-haul truck driving industry is going to), what does this mean for the way that we consume entertainment. Or rather, on who produces the entertainment?
Netflix & Spotify | Music & Movies
Movies, music and stories would fall to the AIs, as they would be able to flood the market with content (hyper-personalized, if you can pay for the higher-end subscription). Generating these movies is likely to be cheaper than producing this content the old-fashioned way, and so traditional studios are likely to die off or be forced to adapt. The streaming wars have shown that traditional studios aren’t nimble enough to adapt unless they have a huge catalog of material. That’s where the likes of Disney & WarnerMusic, etc. will hold an advantage as they would have access to training data like no one else. Just like the streaming wars, this is a war that’s coming in the next 15-20 years. And just like the streaming wars, the first-mover advantage (and associated patents) are going to be the battleground for the next generation of entertainment companies (or Netflix and Spotify if they look at a 20-year long horizon)
Imagine a TikTok that instead of curating videos based on AI is generating videos based on AI.
As far as I can imagine, sports might remain the only bastion of entertainment that is produced by humans in this near future of ours. We have had computers / robots beating humans at games for a fair few decades now (most famously Chess, and more recently famously DOTA and Go). None of this has resulted in a waning interest on the part of the customer for these forms of entertainment. In fact, with the advent of Twitch, watching real humans play these games is one of the fastest growing forms of entertainment. This is likely to survive the post-AI world.
The other set of sporting activities are the traditional ones – the ones where physicality is involved – basketball, Olympics, etc. While the operations of these companies might get streamlined using AI, human tribalism might still lead to these sports surviving as one of the last forms of entertainment produced by humans.
But the popularity of fantasy leagues would suggest that there might be a small but not insignificant appetite for completely AI-generated sporting leagues as well (especially if you don’t even have to simulate humans, and can just simulate marbles).
Addressing The Counterpoint
When normal articles began to be mass manufactured, there came a luxury segment of hand-made objects (cars by Ferrarri, bespoke suits, etc.) Something similar would happen to ‘art’.
Most people would only be able to afford to consumer AI-generated media, but the rich echelons of society would consume the same or ‘better’ media performed by artists in theaters, live music events, etc. This could happen. However, if the ‘human-art’ being performed isn’t competing with the standards of AI-generated art, then all the human performances would be of AI-generated art, at which point, they become no better than cover-bands.
So, in my brief thought experiment, the only entertainment industries that might survive the oncoming AI onslaught would be the the sporting and e-sporting industry and therefore companies like Netflix and Spotify are in the movies & music space, who have been trying very hard to pivot to own-content and decreasing production costs could do well to take a look very deeply at accelerating these technologies to ensure a first-mover advantage.
I am fascinated by this subject. If you’ve come across some good articles, or have thoughts of your own, please do share them with me.