The Creative Opportunities from Gen AI
Creativity has always drawn on new tools and technology. In the Renaissance period, the development of mathematical techniques for linear perspective unlocked new types of more dimensional art. Lithography, invented in the 18th century, enabled mass production and the democratisation of pictures, while the 19th century's camera created whole new forms in photography and later cinema.
When we think of industries transformed by the personal computer, it's often easy to default to accounting and the spreadsheet - but in practice, the impact of the digital revolution on the creative sector was, if anything, even more significant. The computer enabled Photoshop and nonlinear editing, the CGI revolution in special effects, the democratisation of distribution through platforms like YouTube, and entirely new mediums, from video games to podcasts.
Today, the generative AI revolution looks set to be as, or more, significant for the creative sector. Gen AI can:
Write prose, poetry or lyrics across languages or genres;
Create photorealistic images - and increasingly, video;
Create new songs entirely from scratch;
Produce 3D models or interactive worlds;
In order to understand its impact more, we produced new modelling looking at some of the opportunities enabled by gen AI, as well as trying to understand better what its impact could be on employment.
Gen AI media could enable several key benefits for the creative sector:
Help save time on routine tasks. In the last few years, the negative impacts of overwork and crunch culture in fields such as gaming or post production have been increasingly recognised. Despite soaring budgets and headcounts, crews often find themselves working unhealthy hours to meet a deadline. Gen AI tools can help automate more routine tasks such as rotoscoping or transcription, freeing up time for other tasks. On average, we estimate that gen media tools could boost the productivity of individual creators globally by 20%, saving them the equivalent of 390 hours a year.
Make it possible to reach a wider audience. Gen AI tools can make it easier it recut or remix existing material for different audiences, formats and lengths, helping it travel further. One of the most powerful instances of this can be in translation. In total, we estimate that AI-based translation could increase the reach for non-English speaking creators globally by over 350%. That could help generate a further $50 billion in revenue.
Make it easier to prototype new ideas, or enable entirely new types of art. The most powerful impact of gen AI is likely to be not just helping us make today's art quicker, but enabling entirely new types of form. Like any new transformative technology, this is the hardest to predict - but already we are seeing new ideas like:
Artist Refik Anadol using AI to interpret and transform more than 200 years of art at MoMA
Nike using AI to create custom shoes for athletes ahead of the Paris Olympics
New types of responsive interaction fiction, such as AI Dungeon
In total, we estimate that the the overall economic potential of generative media for the core creative sector globally could be over $100 billion by 2035.
This could be particularly powerful for countries like India, where despite Bollywood it lags behind the traditional creative superpowers, such as the US, UK, and Japan. The overall economic potential of generative media for the core creative sector in India could be over INR 220 billion ($2.7 billion) by 2035, while AI-based translation could help increase the reach for Hindi-speaking creators by over 280%.
What does this mean for the existing workforce?
In 2023, former Disney Studios chair Jeffrey Katzenberg famously predicted that within 3 years, the number of animators needed for a blockbuster film would fall by 90%. In practice, last years biggest film - animated or otherwise - Zootopia 2, had a budget of over $150 million, helping support a crew of around 700 people.
What will the actual impact of generative AI media tools be on people who already work in the creative sector?
It's easy to jump to the assumption that a productivity increase of 20% means that employment will similarly fall by 20%. But in practice, there are several reasons to think this is unlikely - and our own modelling instead projects continued employment growth.
To start, while AI can be significant augmenter of capabilities, there are relatively few occupations in the creative sector where it can perform all the tasks that make up each job. For example, a task might require human dexterity in the physical world, creative breakthroughs or consistency over an extended period of time - there any many types of work for which AI is not quite yet ready.
The more significant use of greater productivity is likely to be in meeting higher demand, rather than cutting jobs. To start with, most outputs of the creative sector as leisure goods tend to have a high income elasticity of demand: that is, the richer people get, the more entertainment they want to consume. This is likely to be more important as the world economy keeps growing.
Secondly, the general pattern of productivity developments driven by technology in the entertainment sector is clearly not that they reduce budgets and the workforce. Instead, we more than use up our new technological abilities in higher production values: creating ever more elaborate worlds for us to experience and explore. In the 1990s, a Hollywood blockbuster at the cutting edge of special effects like Jurassic Park might have 50 shots with CGI in them. Today, a film in the Avatar series can have over 1500, and each one is far more elaborate. According to its director Byron Howard, Zootopia 2 was "the biggest movie we’ve ever done in the history of Disney animation, which is saying something... We’ve got one shot that features 50,000 animals on the screen at the same time!”.
Thirdly, greater capabilities don’t just augment those already in the creative sector, but allow new people to enter. New technologies make different types of business model, format and medium possible, enabling the creation of new genres of content. In the early 1990s, there was nothing really equivalent to today’s YouTube creator - and formats like TV or movies were constrained to relatively standardised lengths and formats, rather than the explosion of ultra short or ultra long we see today. Similarly, gen media is likely to make it viable to monetise new types of content that just wouldn’t be feasible to look at today.
Another way to look at this is through a thought experiment where instead of looking forward, you look back. Imagine in the early 1990s you were told of all the new technologies the digital revolution would bring with it: Photoshop, non linear editing, CGI effects, and so on. What would you have predicted would happen to the people employed in the creative sectors then?
In order to explore this more, we created a second toy model where using as similar task based O*NET methodology as we could to those often used to estimate the future impact of AI - but instead looking at a hypothetical impact on digital revolution.
While we can't do this perfectly given data limitations, we found a significant divergence. In our modelling, if you looked at productivity impacts alone, not taking into account the demand effects we go through above, you’d have predicted jobs in 16 core creative occupations to nearly half between 2004 and 2023. In practice, employment instead grew by over 50% in total, and across 12 of the 16 occupations. The number of people employed in animation and special effects has fallen - but by under 1%, way under any reasonable estimate of productivity change. Higher demand and improving the consumer experience more than outweighed productivity changes, with globalisation and offshoring clearly more significant drags on employment than technology.
Nobody can know for sure the future impacts of any technology. But based on what we know now, it seems likely that gen media technologies will help create far more jobs than they remove.
