By Rohana Rezel

OpenAI, the company most responsible for convincing the world that human creativity is obsolete, has been quietly hiring human writers, designers, and editors at salaries north of $200,000.^[1]https://www.linkedin.com/jobs/view/copywriter-creative-studio-at-openai-4344198805/ Meanwhile it shut down its Sora consumer app, the much-hyped AI video product it launched to great fanfare.^[2]https://www.hollywoodreporter.com/business/digital/openai-shutting-down-sora-ai-video-app-1236546187/ You’d think the contradictions would prompt some reflection. They haven’t.

The tech boosters keep insisting AI-generated content taking over the world is inevitable. The data says otherwise.

Nobody actually likes this

A new NBC News poll found just 5 percent of Americans have a very positive view of AI. Another 21 percent are somewhat positive. On the other side: 24 percent somewhat negative, 22 percent very negative, and 27 percent stuck in the uneasy middle.^[3]https://www.documentcloud.org/documents/27777984-nbc-news-march-2026-poll-03-08-2024-release-final/ Those numbers don’t match the media coverage, which reads like a sustained celebration.

The skepticism has been building as the technology moves from demo reel to daily reality, and the daily reality keeps disappointing.

The productivity problem

For two years the central promise of AI in the workplace has been productivity. Work less, accomplish more. Let the machines handle the tedious parts. A study of 163,000 workers tracked by employee-monitoring software found time in deep, focused work is down nine percent, while email and messaging have doubled.^[4]https://www.wsj.com/tech/ai/ai-isnt-lightening-workloads-its-making-them-more-intense-e417dd2c

Research in the Harvard Business Review found 26 percent of content workers are experiencing “brain fry,” a form of cognitive overload from constantly managing, correcting, and second-guessing AI outputs instead of just doing the work.^[5]https://hbr.org/2026/03/when-using-ai-leads-to-brain-fry Another study found 41 percent of workers have been sent AI-generated “work slop” so bad it had to be redone from scratch, taking two hours on average.^[6]https://hbr.org/2025/09/ai-generated-workslop-is-destroying-productivity

None of this proves AI can’t improve productivity under the right conditions. It does suggest most enterprise rollouts are not those conditions.

The ROI that wasn’t

A survey of 6,000 executives for the National Bureau of Economic Research found 91 percent reported no measurable impact on productivity or employment. The average gain: 0.2 percent.^[7]https://www.nber.org/system/files/working_papers/w34836/w34836.pdf

These are the people who spent hundreds of billions deploying AI tools across their organizations. Not academics, not skeptics. The buyers. Most of them got nothing back.

The hype holds anyway

None of this has slowed the hype cycle. The media coverage stays relentlessly optimistic, vendors keep attaching “AI-powered” to every product, and the venture capital hasn’t cooled.

What’s actually changing is what workers are reporting: creative staff told their jobs were disappearing, now spending hours fixing AI drafts that need more work than starting fresh would. Managers paying for subscriptions that haven’t moved the needle. Knowledge workers doing cleanup on content that should have taken minutes to generate.

OpenAI still needs human writers at $200,000 a year. Worth sitting with that.

Acknowledgement: This article was inspired by a Mike Rosenberg LinkedIn post^[8]https://www.linkedin.com/feed/update/urn:li:activity:7442670525987983360/.

Rohana Rezel is a technologist, researcher, and community leader based in Vancouver, BC.

Discuss this article on boreal.social