In what ways will generative AI alter video platforms?

An artificial intelligence model called generative AI is able to produce text, graphics, audio, and video by foreseeing the next word or pixel after being trained on enormous datasets. A text description is the most basic input (sometimes referred to as a prompt) for generative AI. A text-to-image model like Stable Diffusion, a generative pre-trained transformer (GPT), MusicLM, and Imagen Video may all produce audio and visual content based on a written description. This technology will make all forms of content production more accessible. It may equalize the playing field for video production more than cellphones and social video platforms currently have. The video content sector will likewise undergo a profound upheaval as a result.

Consider the leaders in this industry: Netflix, TikTok, and YouTube. These three platforms work by rewarding content providers to create engaging material, connecting the appropriate content to the right consumer, and understanding what content drives engagement, despite the fact that each is distinct in terms of content type and economic model.

An artificial intelligence model called generative AI is able to produce text, graphics, audio, and video by foreseeing the next word or pixel after being trained on enormous datasets. A text description is the most basic input (sometimes referred to as a prompt) for generative AI. A text-to-image model like Stable Diffusion, a generative pre-trained transformer (GPT), MusicLM, and Imagen Video may all produce audio and visual content based on a written description. This technology will make all forms of content production more accessible. It may equalize the playing field for video production more than cellphones and social video platforms currently have. The video content sector will likewise undergo a profound upheaval as a result.

Consider the leaders in this industry: Netflix, TikTok, and YouTube. These three platforms work by rewarding content providers to create engaging material, connecting the appropriate content to the right consumer, and understanding what content drives engagement, despite the fact that each is distinct in terms of content type and economic model.

One of these factors reinforces the others to generate a flywheel that has aided all three platforms in rapidly gaining viewers. Yet the flywheel is starting to slow down. By establishing a new value chain for the production of video content, generative AI will exacerbate their issues.

Why YouTube, Tiktok, and Netflix are having problems

Due to their capacity to assess the relevance and engagement of the material, Netflix, TikTok, and YouTube have prospered. Large quantities of information on who watches what and how are available to them all. Despite their achievements, there are still two significant obstacles to establishing the "what":

Obtaining practical, accurate characteristics. When a video is commissioned (as it is at Netflix), its genre, cast, runtime, and other characteristics are known. Yet, they are general and sometimes subjective designations, making it challenging for an algorithm to draw any conclusions from them. Naturally, many aspects of the video may be described; the screenplay, shot list, and other aspects of production are well known. Yet, efforts to exploit this data may go too far; there may be too much data to adequately characterize a single movie.
Removing obstacles to creativity. Production of closed, Hollywood-style material is costly and cumbersome. In 2022, Netflix spent $17 billion on programming. But if they give a Wednesday and a Glass Onion every week, said Netflix co-CEO Greg Peters, they'll get the overwhelming majority of those people back. With their present production approach, they can't yet provide a Wednesday (a well-liked and expensive contemporary take on The Addams Family) every week.

The user-generated content production methodology utilized by TikTok and YouTube serves as an alternative. It is generally quick and inexpensive, but it requires creating incentives that balance three (often competing) goals: 1) keeping the important creators; 2) inspiring new artists; and 3) maintaining and expanding the viewership. Platforms in this market are starting incentive battles as they attempt to generate an ample volume of interesting content from a select group of well-known contributors. As one instance, TikTok purportedly uses "heating" to manually boost videos. Whereas TikTok required at least 100,000 followers, YouTube Shorts has dropped the threshold for creators to get paid – they now just need 1,000 subscribers.

These two difficulties contributed to the short-lived streaming network Quibi's downfall. Quibi created a platform that integrated all three of these issues. By bringing in high-priced authors and performers, it strengthened the closed, Hollywood-style production structure. Quibi placed its bets on well-known performers and creators, as opposed to promoting independent producers as YouTube and TikTok have done. It received subpar (possibly second-tier) stuff in exchange, and it wasn't helpful. That's because it focused on Gen Z and Millennials without encouraging artists in those age groups. Surprisingly, it didn't utilize AI to decide what content to create either (although it used AI to recommend viewers what to watch).

These difficulties have not yet been addressed by any human-driven platform. There could, however, be a fix. With the advent of a whole new kind of AI-enabled platform, generative AI will alter what video content is produced, how it is produced, and who sees it.

Toward the generative architecture

One of these factors reinforces the others to generate a flywheel that has aided all three platforms in rapidly gaining viewers. Yet the flywheel is starting to slow down. By establishing a new value chain for the production of video content, generative AI will exacerbate their issues.

Why YouTube, Tiktok, and Netflix are having problems

Due to their capacity to assess the relevance and engagement of the material, Netflix, TikTok, and YouTube have prospered. Large quantities of information on who watches what and how are available to them all. Despite their achievements, there are still two significant obstacles to establishing the "what":

Obtaining practical, accurate characteristics. When a video is commissioned (as it is at Netflix), its genre, cast, runtime, and other characteristics are known. Yet, they are general and sometimes subjective designations, making it challenging for an algorithm to draw any conclusions from them. Naturally, many aspects of the video may be described; the screenplay, shot list, and other aspects of production are well known. Yet, efforts to exploit this data may go too far; there may be too much data to adequately characterize a single movie.
Removing obstacles to creativity. Production of closed, Hollywood-style material is costly and cumbersome. In 2022, Netflix spent $17 billion on programming. But if they give a Wednesday and a Glass Onion every week, said Netflix co-CEO Greg Peters, they'll get the overwhelming majority of those people back. With their present production approach, they can't yet provide a Wednesday (a well-liked and expensive contemporary take on The Addams Family) every week.

The user-generated content production methodology utilized by TikTok and YouTube serves as an alternative. It is generally quick and inexpensive, but it requires creating incentives that balance three (often competing) goals: 1) keeping the important creators; 2) inspiring new artists; and 3) maintaining and expanding the viewership. Platforms in this market are starting incentive battles as they attempt to generate an ample volume of interesting content from a select group of well-known contributors. As one instance, TikTok purportedly uses "heating" to manually boost videos. Whereas TikTok required at least 100,000 followers, YouTube Shorts has dropped the threshold for creators to get paid – they now just need 1,000 subscribers.

These two difficulties contributed to the short-lived streaming network Quibi's downfall. Quibi created a platform that integrated all three of these issues. By bringing in high-priced authors and performers, it strengthened the closed, Hollywood-style production structure. Quibi placed its bets on well-known performers and creators, as opposed to promoting independent producers as YouTube and TikTok have done. It received subpar (possibly second-tier) stuff in exchange, and it wasn't helpful. That's because it focused on Gen Z and Millennials without encouraging artists in those age groups. Surprisingly, it didn't utilize AI to decide what content to create either (although it used AI to recommend viewers what to watch).

These difficulties have not yet been addressed by any human-driven platform. There could, however, be a fix. With the advent of a whole new kind of AI-enabled platform, generative AI will alter what video content is produced, how it is produced, and who sees it.

Toward the generative architecture

Consider this circumstance. An author provides the following text description: At an Art Deco café, two people are seated while the snow is falling outdoors. One of them says, "I'm creatively constipated," and then bites into a piece of Swiss cheese.

Very instantaneously, a sound-added, hyper-realistic live-action video is produced and displayed to billions of people. We are aware of the precise input utilized to produce the video in addition to the number of viewers, length of viewing time, portions skipped, likes, shares, comments, searches, and off-platform conversations about the video. This hypothetical situation resolves the two issues with the current video platforms in a single step. The input text box gives a far more accurate description of the film, and it considerably reduces the obstacles to production (it just requires that you type down your ideas). There's no need to use actors or even CapCut.

Consider this circumstance. An author provides the following text description: At an Art Deco café, two people are seated while the snow is falling outdoors. One of them says, "I'm creatively constipated," and then bites into a piece of Swiss cheese.

Very instantaneously, a sound-added, hyper-realistic live-action video is produced and displayed to billions of people. We are aware of the precise input utilized to produce the video in addition to the number of viewers, length of viewing time, portions skipped, likes, shares, comments, searches, and off-platform conversations about the video. This hypothetical situation resolves the two issues with the current video platforms in a single step. The input text box gives a far more accurate description of the film, and it considerably reduces the obstacles to production (it just requires that you type down your ideas). There's no need to use actors or even CapCut.

Even though it doesn't exist yet, this sounds like magic, but it's really simply a combination of three AI systems. Based on the text input, AI #1 produces the video. AI #2 links the proper viewers with the video. AI #3 makes use of the ensuing interaction to advise artists on their next works. The Seinfeld spoof comedy "Nothing, Forever," which employs generative AI to build the screenplay and has nearly 100,000 followers, is perhaps the best example of a more basic implementation of this production methodology in use today.

By advising producers on what drives engagement and presenting viewers with appropriate material, the generative AI-driven video platform lowers barriers to value generation. The creators' ability to produce value outside of the company is increased as a result of the lowered obstacles and better supervision. Moreover, makers are also viewers and vice versa due to the very nonexistent friction between producing and consuming pertinent material. The line gets even more hazy if the viewer enters in a search and that text then serves as the trigger for a new video.

There will be a significant economic effect. Historically, a platform's tiny fraction of very popular material has compensated for a substantial portion of its less popular content. By providing artists with algorithmic suggestions for what to do next, a generative AI platform will boost the popularity of the most well-liked material. At the same time, the remaining will be more profitable due to the much decreased creation hurdles.

What adjustments will the top incumbent platforms make? Netflix is the one of the three that is most wedded to its business model and will probably have the hardest time making a significant shift. It resisted using an ad-supported model for a very long time and has just lately started to do so. In terms of its business model, capabilities, and flexibility compared to what we anticipate, TikTok is the most similar to a Generative AI Platform; yet, it is now the subject of regulatory investigation in the United States. As a result of its aggressive efforts to compete, including the introduction of Shorts and improved creative incentives, YouTube is in a good position. Moreover, it is supported by Google's AI capabilities. Yet Google has previously shown that it moves slowly when it comes to commercializing generative AI.

Even though it doesn't exist yet, this sounds like magic, but it's really simply a combination of three AI systems. Based on the text input, AI #1 produces the video. AI #2 links the proper viewers with the video. AI #3 makes use of the ensuing interaction to advise artists on their next works. The Seinfeld spoof comedy "Nothing, Forever," which employs generative AI to build the screenplay and has nearly 100,000 followers, is perhaps the best example of a more basic implementation of this production methodology in use today.

By advising producers on what drives engagement and presenting viewers with appropriate material, the generative AI-driven video platform lowers barriers to value generation. The creators' ability to produce value outside of the company is increased as a result of the lowered obstacles and better supervision. Moreover, makers are also viewers and vice versa due to the very nonexistent friction between producing and consuming pertinent material. The line gets even more hazy if the viewer enters in a search and that text then serves as the trigger for a new video.

There will be a significant economic effect. Historically, a platform's tiny fraction of very popular material has compensated for a substantial portion of its less popular content. By providing artists with algorithmic suggestions for what to do next, a generative AI platform will boost the popularity of the most well-liked material. At the same time, the remaining will be more profitable due to the much decreased creation hurdles.

What adjustments will the top incumbent platforms make? Netflix is the one of the three that is most wedded to its business model and will probably have the hardest time making a significant shift. It resisted using an ad-supported model for a very long time and has just lately started to do so. In terms of its business model, capabilities, and flexibility compared to what we anticipate, TikTok is the most similar to a Generative AI Platform; yet, it is now the subject of regulatory investigation in the United States. As a result of its aggressive efforts to compete, including the introduction of Shorts and improved creative incentives, YouTube is in a good position. Moreover, it is supported by Google's AI capabilities. Yet Google has previously shown that it moves slowly when it comes to commercializing generative AI.

What happens after the opening credits

Recent technological advancements in understanding of generative AI have been nothing short of astounding. Indeed, we lack the capability to create hyper-realistic, live-action video from a text input, and having access to this technology is essential to the success of the new platform.

Even when information is accessible, the text inputs may often be inadequate to provide a detailed enough description of the video. When creators and viewers create similar texts, the platform is likely to produce a variety of videos that are similar but not identical. How will the platform handle the conflict of interest between the platform and producers as it learns how to generate interesting content? How will the platform stop unauthorized deep fakes, the propaganda and the misleading information that will inevitably be spread?

What happens after the opening credits

Recent technological advancements in understanding of generative AI have been nothing short of astounding. Indeed, we lack the capability to create hyper-realistic, live-action video from a text input, and having access to this technology is essential to the success of the new platform.

Even when information is accessible, the text inputs may often be inadequate to provide a detailed enough description of the video. When creators and viewers create similar texts, the platform is likely to produce a variety of videos that are similar but not identical. How will the platform handle the conflict of interest between the platform and producers as it learns how to generate interesting content? How will the platform stop unauthorized deep fakes, the propaganda and the misleading information that will inevitably be spread?

Despite these limitations, it is quite conceivable that generative AI will power brand-new video content platforms that replace or at the very least complement the present iterations of Netflix, YouTube, and TikTok. In addition to powering content creation, generative AI will also be utilized to fuel platform dynamics between the platform, content providers, and audience members. It should almost go without saying that none of this is without technical danger and moral ambiguity. Video is just one area where we anticipate to see such quick change, of course. For those who can identify areas that are vulnerable to disruption or who want to use generative AI to secure their position, many other creative fields in literature, music, and the arts are in for significant upheaval and new financial prospects.