Elon Musk just did an extended interview with Peter Diamandis on the Moonshots podcast. It was a 3 hour marathon covering a variety of topics.
Here is the video if anyone wants to see it.
One of the things that stood out to me was Elon's proclamation that Hollywood will be dead within 5 year. This is something I wrote extensively about. To be fair, Musk did not frame it in these exact terms.
What he did say is that on-demand video will be one of the larger drivers of AI by 2030. That is only 4 years from now. Think about that for a second. Of all the promises by AI proponents, Musk is saying that on-demand video will be the largest use case.
If true, that boggles the mind. We know how much chatbots are being used. Of course, these are still text based. Video generation is improving. The improvement over the last couple years was significant.
There is a lot of conversation about the troubles Hollywood is facing. It has penetrated the mainstream consciousness with the financial and traditional networks picking up the story.
Over the last few years, the box office numbers were pathetic. The mantra for the industry was "wait until 2025". Last year was a dud, barely edging out a gain from a poor 2024. The challenge is the number of releases far exceeded the previous year. Without that, another decline was in the cards.
The reason people give is all over the place. Ticket prices, poor scripts, out-of-touch actors, and woke are all put forth. While they likely had a hand in things, none were the overriding factor.
We are simply looking at a technology change. It started with the breaking of the monopoly of content distribution and is not spreading to the creation. Generative AI is still in its infancy yet the improvements are rather impressive.
Here is a trailer for the movie Zelda that was created using AI.
This is a big step up from the Will Smith eating spaghetti video of a few years ago.
AI will dwarf the impact of the internet.
Consider what had to be done to bring that to the masses. For billions of people to access the internet, it had to be built. This meant laying billions of millions of cable, setting up cellular towers, and building the cloud.
Today, that acts as the distribution system for AI.
Again, ponder the impact of this. Anyone can download AI (or access it via a website). We could not do that with electricity, the steam engine, or automobiles. Those had to be delivered in a physical form.
We have more than 5 billion people with phones that access the internet. These people are gaining access to more AI services on a daily basis.
Therefore, Musk's prediction makes sense. Many are overlooking the impact due to the fact that AI is still lagging. The fact it cannot create a feature length film of cinema quality today should not dissuade people. It is only a matter of time.
The advantage Musk has is the fact that he is deeply involved in xAI. He is seeing where the next generation models stand. The improvement from each version is where the exponential projections are.
Many estimate that we are seeing a 10x in AI capabilities each year. This is due to advancements in algorithms, GPUs, and other aspects of the stack. Each is having a rate of improvement that compounds when spread across the entire stack.
The cost per token keeps declining. Models are showing up offering the capability of the larger models yet running locally. There is a 6 month lag but considering the money invested by Big Tech for the LLMs, this is really eye opening.
Speed is key. Things are not slowing down. Moore's Law is dead since it is being dwarfed. Just think about the difference between a doubling every 2 years versus a 10x yearly.
We are talking orders of magnitude more.
