On 4 May 1948, the US Supreme Court ruled in a landmark antitrust case that would change the way that Hollywood studios worked forever. Until that point the studios had owned everything, from the production lots, through to the cinemas that the films played in. Actors, Directors and writers were tied in to long-term, exclusive contracts that prevented them from working with any rival studio. If you were an actor on a contract, your studio decided how you looked, what films you did, and even what your screen name was. The talent was, in effect, another asset on the balance sheet.
The Supreme Court ruling was the beginning of the end for the monopolistic, vertically integrated studio system. The ‘Hollwood Model’ that emerged separated production from distribution, ended long-term, exclusive contracts, and allowed much greater creative freedom and agency for actors and behind-the-camera filmmakers. A growing collection of independent producers and studios could now make films free of major studio interference. Films became temporary organisations, brought together for one production and then disbanded when it wrapped. A unique combination of talent was assembled for each film project. Reputation no longer sat exclusively with the major studios but with individual talent. The work had effectively become unbundled from the company whose logo was on the credits.
The ‘Hollywood model’ is being held up by some as a future pattern for a fluid, project-based world of work in the age of AI. A world where independent talent moves between ‘productions’ or projects and reputation travels with the individual and their legacy of work. We’re already seeing how jobs and roles are being unbundled by AI into a collection of tasks, some of which can be automated, and some of which can’t. Jobs that have specific tasks which are less dependent on other parts of the job can more easily be unbundled and so are more likely to be heavily impacted. Jobs where there is a high coordination cost to unbundling (the market buys all the interdependent parts of the job as a bundle) will stay human. But what if that logic is applied to organisations as well? What if the company is being unbundled by AI into a series of projects? Once work becomes readable as a set of discrete tasks does the weight of logic still favour these tasks being addressed primarily by permanent teams and FTEs? Ronald Coase famously defined the firm through its ability to bypass the transaction costs of the open market but AI is collapsing the costs of coordinating, specifying and verifying work within and across team and organisational boundaries. When those coordination costs fall far enough, the cost/benefit of assembling a bespoke team per project may beat maintaining a permanent team and the Hollywood model could well be what we get.
It’s easy to assume that this represents a somewhat chaotic and unstable future world of work. But as Beth Bechky’s ethnography of film crews has shown there are some consistent practices which enable the whole thing to work. Film crews are organised around structured role systems. These roles can be applied to organise both immediate work and to maintain continuity across different projects. Even if they’ve never worked together before a Director can work with a cinematographer from day one because there is an expectation around the role that each will play, and nuances around that can be sorted out project by project. Beth also notes how this kind of structure enables film crews (and also SWAT teams whom she studied) to handle unexpected events through ‘organisational bricolage’, restructuring their activities on the fly by shifting roles, reorganising routines and reassembling the work. The bricolage depends, she says, on sociocognitive resources built from shared agreement and interactions, notably building cross-member expertise so that people understand each other’s jobs. In this sense it’s a structure built for collaboration and flexibility.
Fluid talent models are not unique to film production of course. In the 1960s you could buy a record by the Beach Boys, the Byrds or the Monkees and without knowing it actually be listening to the same musicians. A loose-knit but legendary collective of players who later became known as ‘the wrecking crew’ played on hundreds of iconic pop and rock hits through the 60s and 70s. Musicians like Carol Kaye, Hal Blaine and Tommy Tedesco turned up at the studio, played whatever they were asked, and left with a flat fee. The band on the sleeve got all the adulation and the musicians that played much of the music on the recording were left uncredited.
In the case of the wrecking crew, the reputation for the work they carried out was not public, but instead circulated amongst the producers who knew who to call. There were no career ladders to climb. There was a body of work, and a network of people who could hire you again. If AI does unbundle firms into projects and erodes the once relatively fixed boundaries that determined who is in a business and who is not, reputation and network will become significantly more important in the world of work. An interorganisational career progression will require reputation to be portable and visible across projects, role clarity, and a coordination layer that can make it workable.
The decoupling effect of AI raises some interesting questions. A ‘job’ was always a relatively arbitrary bundle of tasks that were aggregated together out of coordination convenience. So if AI unbundles jobs into tasks what even is a ‘job’ anymore? If an organisation’s output was generated by a relatively stable collection of people performing quite narrowly defined roles but those tasks are now much more widely distributed, what even is a ‘company’ anymore?
And these are not the only forms of decoupling that we’re likely to see. AI is increasingly disrupting the historical correlation between time/effort invested and the output produced. An employee might describe to their boss how it took three weeks of work to produce an output but for many forms of knowledge work the effort to output ratio is now collapsing, or at least changing quite radically. Sangeet Paul Choudary recently made the point that our habit for measuring work by outputs has distorted our perception of value, leading to an overvaluation of the artefacts of work (presentations, spreadsheets, reports) and an under-appreciation of the challenge that those artefacts are attempting to address:
‘When output becomes abundant, a polished artifact isn’t as valuable anymore…What’s more valuable is understanding what job needs to be done and what form the idea should take to do that job well.’
In the age of AI, says Sangeet, value will increasingly shift to our ability to break ideas and/or challenges into components that can be brought together in specific ways to generate the right output for the right context (as an example he compares his book, which is a group of ideas bundled in a specific way for a particular need, with a concept map that he has created which enables people to explore individual ideas from the book and see how they are linked together).
The collapsing cost of producing competent outputs in work is also placing far more emphasis on discernment and discrimination – knowing what’s needed, what’s worth doing, and whether an artefact is any good or not. Execution is being unbundled from judgment. So rather than execution and artefacts being central to how we might once have proved our abilities and gained our reputation, the ability to demonstrate good judgement and discernment from abundance becomes key. It’s no accident that the freelance market is bifurcating, with those able to prove good judgment, creativity, and imagination commanding premiums over those that are focused on now commoditised forms of execution and delivery.
The longer term issue here of course, is that good judgment has often been developed through years of delivery and execution. And if, as happened with the wrecking crew, getting paid depends on your network and relational reputation as well as your portfolio of work, how will juniors ever be able to develop their networks without a consistent succession of projects and different people to work with? There’s a lot of discussion already around how skill acquisition is at risk of being decoupled from doing the work, with AI threatening junior level tasks that used to be the training ground. But the subtler risk here in that AI is also threatening proximity, and the ability for juniors to learn and build tacit knowledge through working alongside someone that is much more experienced than they are.
At the other end of the scale AI is also decoupling seniority from managing people and capability from headcount. AI will absorb the coordination and information-routing work that once justified layers of middle management, meaning that seniority is no longer as directly tied to the number of people that you manage. As an example, the CEO of Coinbase recently announced that he was flattening the org chart to a maximum of five layers and that senior staff were expected to play a ‘player-coach’ role where they will be work do-ers and not just work coordinators and facilitators. With AI, team and organisational capacity no longer scales linearly with the number of people, meaning that headcount will no longer be seen as a measure of capability or status. Once that link is broken we get all kinds of strange effects like tiny firms operating at an enterprise scale, or individuals who are able to generate disproportionate value. It’s going to be interesting to watch that one play out.
And what about identity being unbundled from occupations? Our occupation has for a long time been a primary source of meaning for us. So when a stable job increasingly dissolves into a shifting portfolio of projects and roles, our identity is much less wrapped up in our job title. This is both a potentially freeing but also a potentially disorienting change. It certainly makes it increasingly difficult to efficiently answer the eternal question ‘what do you do?’ (as I have found to my own cost).
AI is decoupling so many things that organisations in the industrial era had welded together for efficiency. Jobs and tasks. Output and effort. Judgement and execution. Seniority and management. Skills and doing. Reputation and employer. AI is like a ubiquitous solvent disolving once tightly defined bonds between elements that have defined the world of work for decades. It’s going to be fascinating to see not only what replaces those bonds, but also how this impacts the fundamental relationship that workers have with who pays them for their contributions.
A version of this post appeared on my weekly Substack of AI and digital trends, and transformation insights. To join our community of over thirteen thousand subscribers you can sign up to that here.
To get posts like this delivered straight to your inbox, drop your email into the box below.
Leave a Reply