The Deployment Gap: Where AI Could Transform Entertainment Production — But Isn't Yet

By Jacob Hawley, CEO — Monster Gaming

On March 5, 2026, Anthropic's research team published something that changed how I think about this industry.

Maxim Massenkoff and Peter McCrory — two economists at Anthropic — built a framework for measuring not what AI could do to jobs, but what it's actually doing. They called it "observed exposure." The distinction matters more than it sounds, because every headline about AI replacing workers is based on theoretical capability. Massenkoff and McCrory measured reality. And reality looks very different from the headlines.

Their key finding: AI is far from reaching its theoretical capability. Actual coverage remains a fraction of what's feasible.

That gap — between what AI could do and what anyone is actually deploying — is the most important number in our industry right now. And nobody had measured it for games, film, VFX, or entertainment production.

So I did.

The Entertainment Production AI Exposure Index

I've been building game studios for thirty years. Four studios. 58 shipped games. 85 MobyGames credits. BioShock. Borderlands. Seven years as Worldwide Director of Technology at 2K Publishing, managing a $40M+ annual P&L and delivering 21 AAA titles. I've held or supervised nearly every production role in the industry.

When I read the Anthropic paper, I recognized the framework immediately. It was the measurement layer our industry has been missing as AI reshapes every pipeline from QA to rotoscoping to post-production.

I adapted their methodology — their three-layer framework of theoretical capability, observed usage, and labor market outcomes — from the entire US economy to the entertainment production economy specifically. The result:

• 81 roles across 8 sectors: games, film, TV, visual effects, audio, virtual production, motion graphics, and entertainment business
• ~2,000 discrete tasks, each scored on a 5-point scale for AI theoretical capability
• Observed exposure measured from real production data, public adoption signals, and studio engagement ground truth
• The gap map — where theoretical capability vastly exceeds actual deployment

The methodology, the data, and the taxonomy are MIT-licensed. Open. Because the best era of game development was the one where everyone shared what they learned.

What the Data Shows

Three numbers matter.

0.63 — the deployment gap for game QA. AI can automate the vast majority of QA tasks today. Automated regression testing, visual comparison, platform compliance checking, test case generation — the technology works. But observed adoption across the industry is under 10%. That gap represents the single largest deployment opportunity in game development, affecting 32,500+ QA engineers in the US alone.

0.65 — the deployment gap for assistant editing in film and TV post-production. Media management, transcode workflows, bin organization, sync, VFX pulls — the tasks that consume 60-70% of an assistant editor's day. AI handles all of them. Almost nobody is deploying it at pipeline scale.

0.60 — the deployment gap for VFX rotoscoping and paint. AI roto is production-ready. SAM, DINO, and their successors can pull mattes that took artists days in hours. Adoption is accelerating but pipeline integration — actually getting the AI output into the Nuke comp tree reliably — remains the bottleneck.

These three roles alone represent over 100,000 workers across games, film, and VFX.

The Part Nobody Is Talking About

Here's what the gap map actually tells us — and it's not the story you're expecting.

The entertainment industry has lost 45,000 jobs since 2022. Every headline connects those losses to AI. But Massenkoff and McCrory found something important in their economy-wide analysis: no systematic increase in unemployment for highly exposed workers since late 2022. The displacement narrative is running ahead of the evidence.

What the gap map shows instead is an augmentation window. AI capability exists. Deployment is early. The industry has a choice — right now, in this window — between two futures:

Future A: Wait until AI deployment happens to the industry rather than for it. Displacement becomes reactive. Workers are laid off because competitors adopted first and changed the cost structure.

Future B: Deploy AI proactively as a force multiplier for existing talent. The same QA team covers 30% more surface area. The same VFX pipeline produces 30% more shots. The same post-production house handles 30% more projects. Revenue grows. Headcount stays. Fulfillment increases because people spend less time on mechanical tasks and more time on creative judgment.

Future B has a number attached to it: 1.3x.

The 1.3x Thesis

If the average entertainment production role has a deployment gap of 0.45 — and ours range from 0.13 for on-set physical production to 0.63 for game QA — then closing that gap through AI deployment yields a conservative productivity uplift of 25-35%. Call it 30%. Call it 1.3x.

For EA, at ~$7.5 billion in revenue, 1.3x means $2.25 billion in incremental capacity from existing talent.

For Take-Two, at ~$5.6 billion, 1.3x means $1.68 billion.

For Ubisoft's 19,000 employees — the largest headcount of any Western publisher — the gap map shows where those 19,000 people could produce like 25,000 without hiring 6,000 more.

This is not a cost reduction story. This is a performance story. The studios that deploy AI into the gap become 1.3x more capable without losing a single person who makes their games, their films, or their shows distinctive.

Andrew, Strauss — you don't need fewer people.

You need every creator you have in a Formula 1 car. Teach them how to drive it. Watch how much more fulfilled your teams are and how much happier your shareholders get.

How We Built This

The methodology is borrowed from the best and adapted for our industry.

Layer 1 — Task Taxonomy. We decomposed 81 entertainment production roles into ~2,000 discrete tasks, sourced from O*NET where applicable, supplemented by IATSE/DGA/WGA job classifications, VFX studio role specs, and three decades of production experience across 58 shipped titles.

Layer 2 — Theoretical Capability (β). Each task is scored on a 5-point scale: Can AI perform this task end-to-end (1.0)? With specialized tools (0.75)? Meaningfully assist (0.5)? Marginally help (0.25)? Or not at all (0)? Scored by our production team and calibrated against Eloundou et al. (2023), GDC survey data, and VFX industry benchmarks.

Layer 3 — Observed Exposure. What's actually being deployed in professional production contexts? Measured from our own platform telemetry, public adoption signals (GDC surveys, job postings, freelancer rates, GitHub activity), and ground-truth data from studio engagements.

The Gap. Theoretical minus observed. The roles where this gap is widest are where the deployment opportunity — and the augmentation potential — is greatest.

We score engine-specifically, because a Godot studio has a structurally different gap map than an Unreal studio. Godot's text-based file formats are directly parseable by LLMs. Unreal's binary assets require intermediary tooling. This matters for deployment planning.

What Comes Next

The Entertainment Production AI Exposure Index publishes quarterly. The methodology paper, task taxonomy, and Sector 1 data (game development — 16 roles, 260 tasks, fully scored) are available now under MIT license.

We'll expand to full task-level detail across all 8 sectors over the coming quarters. Each update includes refreshed observed exposure data, labor market signals (job postings, union reports, freelancer rates), and the gap map showing where the deployment window is opening, closing, or stable.

We're also tracking what Massenkoff and McCrory identified as the most concerning early signal: a slowdown in hiring of workers aged 22-25 in exposed occupations. If AI is automating the entry-level tasks that historically served as training grounds, the talent pipeline for the next generation of entertainment professionals is at risk. This deserves industry attention, and we intend to provide the data.

Acknowledgments

This work would not exist without the methodological foundation established by Maxim Massenkoff and Peter McCrory at Anthropic. Their paper "Labor market impacts of AI: A new measure and early evidence" provided the analytical framework. The Anthropic Economic Index team — Ruth Appel, Kunal Handa, Ryan Heller, Alex Tamkin, and their colleagues — built the usage data infrastructure that makes observed exposure measurement possible. Tyna Eloundou, Sam Manning, Pamela Mishkin, and Daniel Rock established the theoretical capability baseline in 2023 that both the Anthropic paper and this index build upon.

They provided the framework. This is my attempt to make it useful for the industries I've spent my career in.

Any errors in adaptation are mine.

Monster Gaming is an AI-native game development platform. 88+ specialized agents. Open protocols. MIT-licensed tools. The journey is the product.

The full index methodology, task taxonomy, and data are available at monstergaming.ai/research.

Contact: info@monstergaming.ai