AI made our ads. We (humans) made them good. Here’s the difference.
Tim Murray
Sr. Creative Director, Brand | Jan 23, 2026
The boom in generative AI is putting a lot of thoughtful professionals in a bind — and adding to the pressure of anyone looking for a new job.
There's the C-Suite demand to squeeze more productivity out of every headcount. There's the icky feeling that AI builds fake worlds from the work of real, uncompensated artists. And there's the fear that if you don't ride the rush of progress, you'll be left behind. But is AI even capable of squeezing out anything but slop? As the brand director of a tech company, I live in the middle of these forces every day.
So, we decided to stop speculating and start making. Our CMO challenged our 10-person internal creative team to produce a series of YouTube spots end-to-end using generative AI. We had two months, a shoestring budget, and a healthy dose of skepticism. It was hackathon time.
We didn't run our experiment to replace a crew. We did it to see if a small team of seasoned creators could harness AI tools to make better advertising on a low budget than we could have without them.
Here is the unvarnished field guide to what we learned, the tools we broke, and why the "AI-will-take-your-job" narrative misses the point.
It’s a pipeline, not a magic button.
If you’re a job seeker worried that a prompt replaces a profession, come off the ledge. AI didn't do the work. It was more like the engine in a car we still had to drive, steer, and occasionally push uphill.
- Strategy: We built a custom Google Gemini agent to act as a strategic sparring partner.
- Visual development: We used Midjourney and Nanobanana to build style anchors.
- Motion: Runway was our workhorse, with Sora and Kling providing specific surreal clips.
- Sound: ElevenLabs for voiceovers (often using our own staff's voices as inspiration).
- The final stitching: We still spent hours in Premiere and Final Cut Pro.
The lesson: AI doesn’t replace the creative process; it’s a shiny new tool in the box that can both save time and waste time, depending on how it’s used.
Discernment is the new hard skill.
Our first attempt was a womp-womp moment. We asked Gemini for a YouTube ad about job-seeker frustration. It gave us a guy at a desk looking worriedly at a laptop. Boring and predictable, it didn’t even rise to the level of slop.
We realized that AI has no lived experience. It doesn't know what it feels like to be ghosted after a third-round interview. It doesn’t have layers of exposure to art and culture that allow human brains to connect dots with weirdly satisfying results. We had to provide the metaphora: What if job hunting was like a high-stakes dating show? What if every job was a door in an endless, surreal hallway? Prompting lit up our brains in exactly the same way it does solving any creative puzzle, and this felt super satisfying.
For professionals: Your value is no longer in just making the thing. Your value comes in defining the right metaphor and making the right judgment calls. The life you’ve lived finds its way into the prompts, and it can’t be replaced.
Our 7 rules for gen AI production
We went in with a guideline for experimenting with short-form video:
- Don’t be boring. Hit them in the first 3 seconds and pace it fast.
- Make one point only. That’s all you have time for.
- Be weird. AI excels at the surreal; don’t ask it to do normal.
- Lean into humor. It keeps the audience on your side.
- Build for sound-off. Your viewer is on the train.
- In this dance, the human always leads. Don’t accept mediocre output.
- Quantity leads to quality. Be prepared to generate hundreds of bad early drafts.
Avoiding the Uncanny Valley
Human faces are the hardest thing for AI to get right. Eyes darted wildly, smiles melted. We found two workarounds:
- Motion references: We recorded our own faces on our phones and fed them into Runway to guide the AI. It needed a human skeleton to build upon.
- Surrealism over realism: We found that the more human we tried to make characters look, the cringier they got. But when we used animals, claymation styles, or surreal environments, our senses were more forgiving.
The lesson: Replacing people with AI is not the solution. Use it instead to build upon your own creative framework — and team.
5 truths for the future workforce
Whether you’re a creative director or a job seeker in any field, we offer these takeaways from our stint in AI School:
- Generative AI is drudgery. Prompting is like cranking a slot machine for eight hours a day, praying for a jackpot, but as with any craftsmanship, labor is required to achieve great results.
- Yes, it saves money, but… AI helped us stretch our budget, but we paid in hours — lots of iteration, lots of troubleshooting, lots of manual cleanup.
- Teamwork triumphs. Our best work came from passing your drafts back and forth, tapping into everyone’s unique skills. Work together in person if you can.
- Human creativity is irreplaceable. As of January 2026, the models cannot produce greatness on their own. While AI is evolving by the minute, it seems unlikely that the lived experience of humans will ever be replaceable.
- Stay curious. People who experiment — even skeptically — will learn how best to work alongside AI by applying the power of human intelligence.
- Invest in your AI self now. You don't need to become an AI expert overnight, but it helps to understand where these tools help and where they fall flat. That's the fluency employers will actually seek.
We didn’t leave this project feeling replaceable; we left satisfied that human insight is indispensable to achieve good results with AI. Dancing at the edge of what’s new can feel scary, but it’s better than sitting out the song.
Did we succeed? Here are the spots for your consideration:
One final note: With generative AI, you need to wear all the hats from talent scout to set designer to foley artist, so our list of credits comes without position descriptions. Thanks to
Mara Lecocq, Becky May, Laura Graham, Amber Lee, Salma Teran, Yuliya Tverdokhlib, Adrienne Michitsch, and Doug Muise. Special gratitude to Miles Gilbert, Benjamin Benichou, Ellen Xu, Will Van Iden at Runway, and the team at Mekanism for their inspiration and guidance, and to CMO Eric Petitt for making our exploration possible.
If you want to follow the play-by-play of this experiment, explore this long Substack piece.
Tim Murray
Tags:AIAI at workCareer AdviceFuture of Workgenerative AI
