143: Mixed bag
Time to mow the lawn, how to pronounce Walz, and some AI infrastructure notes
It’s the first Wednesday of the month in Minnesota which means that sound you hear are the tornado warning sirens being tested. Freelancing is heating up, slightly. This new VW ad is cute, especially considering cupholders were verboten for so many years. Interesting recap here from Vox regarding the failure of so many AI product/service ads to resonate recently, especially during the Olympics. And don’t forget—the Paralympics get started in Paris on August 29.
🎉 Today marks my 8th wedding anniversary AND my 18th year on LinkedIn. Oh joy, oh rapture! Relationships are not without effort. Anthony De Mello’s book, Awareness, has been—without a doubt—the single best guide towards understanding myself in relationship. But venture carefully. De Mello does not pull punches. As he puts it, “You are never in love with anyone. You’re only in love with your prejudiced and hopeful idea of that person.”
🎶 If you’re in Minneapolis this Saturday and appreciate burlesque, the occasional New Orleans street shuffle, the blues, and sound effects mixed with bent lyrics about dogs named Carlos, reprobates, the Devil and unwanted facial hair—have I got a show for you!
🗳️ His last name is pronounced “WALLS.” As in, the walls hold up the building. Or balls, but with a W. And he is indeed a righteous Tim. Yes, I bought the camo hat. Oh. And for all those morons who think #TamponTim is a slight, maybe watch Bodyform’s amazing ad from BBDO AMV. Walz had the guts to sign legislation requiring menstrual products be available in bathrooms of all Minnesota schools for grades 4 to 12.
🤖 The infrastructure for specific, brand-friendly creative use of AI continues to evolve. WPP debuted results of its partnership with NVIDIA at SIGGRAPH last week (here’s NVIDIA’s take). We’re shifting from a hit-or-miss, Wild West approach to generative outputs to results which can be tightly constrained (i.e. a Coca Cola bottle must look like its iconic shape; brand colors must be correct) from the get-go.
This the inevitable outcome of resources and investment. Getting an AI to perfectly represent every element of an image as expected isn’t magic, it is careful, deliberate, expensive work. What WPP are doing is industrializing and standardizing the chaotic, jagged edges. They want to reduce the churn currently inherent in generative creation, to help creatives get to more relevant, accurate and (dare we hope) compelling results faster. They’re making the magic safe for brands with the ability to invest.
🍎🤖 Does prompting still confuse you? Now imagine working at Apple, trying to make the magic of AI as invisible and intuitive (and safe!) as possible. This is another infrastructure story. Apple just released its latest developer betas. As expected, developers dig in…and find things, like the initial hints of how Apple Intelligence might function. In this case, they found pre-prompts for tasks like, “summarize an email,” or “create a video from my photos.” Full coverage from The Verge and MacRumors; more via Reddit.
This is the ugly plumbing which must occur for seamless, intuitive outcomes to “just happen.”
If you or your team need help making sense of all this AI stuff within strategy, creative briefing or ideation and production, I can help. I’m currently delivering 60-minute online courses, and longer format in-person workshops for brands and agencies.
🤔 Conceptually, it is a daring ploy. First telling advertisers, “Don’t advertise. Go f*ck yourself. Is that clear?” Then, sue those same advertisers for not advertising. When we find out Mischief has been secretly coaching Elon the entire time, minds will be blown.
🛠️ Way back before OpenAI, Sam Altman helped write a Startup Playbook for Y Combinator. At 51 pages it is filled with pragmatic but not exhausting entrepreneurial advice. The section on ideas is particularly useful, to wit, “it’s easier to do something new and hard than something derivative and easy.”
🎨🤖 While their tool set isn’t as robust as Midjourney’s or Adobe Firefly’s, I’m impressed with Meta’s evolving ability to generate images. We’ve got three solid generative image-making tools now, and Meta’s is free. While the creative output from OpenAI’s tool trails the others, there is a distinction—if you update your ChatGPT app, you can just talk with the interface and tell it to create an image. Click the headphones icon lower right.
👏🏾🤖 Last but not least: I am so proud of our Minneapolis College of Art and Design students for their courage to engage in generative AI. Shout out to Hewan Dagmawi and Anas Mohamed on completing their summer “creative AI” internship at Target, facilitated by The Brandlab.