
AI
Written with AI from 3 sources, without human review. Model: gpt-5.6-sol.AI and YouTube: Why Reality Still Wins for Creators
AI and YouTube are reshaping video creation, but Mark Rober argues that real experiments, human judgement and audience trust remain creators' strongest defences.
Published by Web2Labs. Report an error in this article
AI can help you make better videos. It can also help 40,000 strangers make your video before lunch.
Mark Rober thinks creators still have a defence: reality. Not polished reality. Actual things happening, with a recognisable human behind them.
Rober's line
Tubefilter reports1 that Rober, whose main YouTube channel has roughly 85 million subscribers, isn't particularly worried about synthetic videos stealing his audience. Watching a computer-generated basketball trick isn't the same as watching somebody risk their teeth doing it. The knowledge that an experiment really happened is part of the entertainment.
But Rober isn't refusing AI. His team uses it while developing scripts and thumbnails. He also sees it helping teachers adapt Class CrunchLabs, his science curriculum for grades three through eight, to different students. Google.org recently gave the project a $2.5 million grant.
So the position isn't "AI bad." It's closer to: let the machine move the chairs, but don't let it host the dinner.
Your moat isn't production quality anymore. Everybody has shiny tools. Your moat is access, judgement and the mildly inconvenient fact that you actually did the work.The real fight
YouTube is playing both sides because, frankly, it has to. CEO Neal Mohan said more than one million channels used YouTube's AI creation tools daily2 in December 2025. In the same January 2026 letter, he named low-quality AI uploads as a problem and said YouTube was adapting its spam and clickbait systems to contain them.
The platform also says repetitive, mass-produced work can lose monetization. Yet using AI itself isn't the offence. Realistic synthetic material needs disclosure, while assistance with ideas, editing or production is generally fine.
TikTok is taking a similar route. It requires labels on realistic AI-generated content3 and has expanded automatic detection. Labels, everywhere. Meanwhile, creators are still complaining that copied formats, synthetic narration and endless template videos are swallowing discovery.
That's the danger for you. Not instant replacement. Attention getting cheaper while trust becomes harder to earn.
Your next move
Use AI backstage. Let it challenge a weak premise, organise research or generate thumbnail variations. Then make the important calls yourself. If the machine can publish the final piece without you, you haven't added enough.
Put evidence on screen. Show the failed attempt, the rough prototype, the awkward conversation, the receipt. Reality has texture. Synthetic content usually sands that texture away.
And build recognisable habits: your voice, your tests, your strange little opinions. Formats will be copied within hours. A relationship with an audience takes longer.
Annoying, yes. Also rather useful.
- 1tubefilter.comTubefilter reports
- 2blog.youtubemore than one million channels used YouTube's AI creation tools daily
- 3newsroom.tiktok.comrequires labels on realistic AI-generated content

Discussion
Join the discussion - share what's working for your channel.