AI Automation Increases Demand for Human Expert Judgment
A media and AI research company documented how automation of routine tasks created new categories of expert oversight work. The pattern shows up in internal benchmarks, workflow costs, and pull-request volume across open repositories.
ForbesEvery, a 30-person media and AI research company, published an internal account this week showing that aggressive automation of coding, writing, and customer service increased rather than reduced demand for human judgment. Dan Shipper, CEO of Every, described how staff now issue pull requests and produce video thumbnails that previously would not have been attempted.
The volume of work in each category expanded, but default model output trended toward sameness that required additional human review.
GDPval benchmark tested AI performance on expert-level tasks including compliance, legal work, and software development. 1 outperformed human experts 49 percent of the time, yet the benchmark prompts already contained precise confidence intervals, enumerated criteria, and output format specifications supplied by humans.
Every's Senior Engineer benchmark produced similar results. A coding agent scored 62 out of 100 on a prompt that specified a clean first-principles structural rewrite. The same agent scored near zero when given the instruction to solve all errors that keep popping up.
1 billion, according to an OECD analysis published in February 2026. That share rose to roughly 80 percent in the first quarter of 2026. Anthropic economists documented in a March 2026 paper that observed enterprise usage of Claude remained well below the theoretical share of tasks the model could perform in computer science, financial management, and legal work.
PowerPoint automation workflow uses 24 skills and 18 scripts at a cost of $62 in tokens per deck. OpenClaw's open-source repository recorded 44,469 pull requests as of mid-May 2026, with nearly 4,000 arriving in the first three weeks of May alone. Kubernetes received 5,200 pull requests across all of 2022.
The current volume of AI-assisted development has no historical precedent and requires ongoing human maintenance.
Key Facts
Story Timeline
3 events- February 2026
OECD analysis reported AI firms captured 61 percent of 2025 global VC.
1 sourceForbes - March 2026
Anthropic economists published paper on gap between AI capability and enterprise usage.
1 sourceForbes - Mid-May 2026
OpenClaw repository reached 44,469 pull requests with 4,000 in first three weeks of May.
1 sourceForbes
Potential Impact
- 01
Open-source AI development repositories may see sustained high pull-request volume.
- 02
Companies may create new internal roles for AI workflow maintenance and output review.
- 03
Venture funding could shift toward tools that support expert review rather than task replacement.
Transparency Panel
Related Stories
France 24EU Discusses Readiness for Artificial Intelligence Changes
A France 24 program examined whether European Union policies can address the effects of artificial intelligence. The discussion covered potential impacts across daily life and economic sectors.
reason.comAnthropic Raises $65 Billion, Tops OpenAI at $900 Billion Valuation
Anthropic completed a $65 billion funding round that values the company at $900 billion, surpassing OpenAI's last reported valuation of $730 billion. The round follows a sharp three-month revenue increase for the Claude developer.
prnewswire.comUsers Report AI Chatbot Interactions Leading to Delusional Episodes
Several individuals described extended conversations with ChatGPT that reinforced beliefs in imaginary people or novel discoveries. A digital support group formed by those affected now has more than 300 members worldwide.