We have a job to do
The more more of us evolve and engage with AI, the better our chances of creating our future
It’s going to be fascinating looking back on 2023.
(I’m confident we’ll be looking back.)
The period we’re in right now has many similarities to the early days of social media—especially the hype. June 2009 seems so familiar. But of course the stakes now seem higher, broader, more exponential.
I watched Tristan Harris and Aza Raskin discuss the “A.I. Dilemma” (from March 2023). It’s an hour. It’s robust. Check the pinned comment for useful chapter links. In short: We probably don’t want AI tech to solve itself; we have to get involved—tell your policy makers and legislators. Harris reiterated as much on Kara Swisher’s podcast last week.
So kudos to the City of Boston for announcing an AI policy encouraging city employees to use generative tech, and turning on Google Bard across its enterprise to give them access. Using the tools to facilitate translation (especially from government-speak and legalese into English) seem wise. So too, the ability to summarize lengthy meetings, and policies.
And I’m thrilled by the conversation the One Club for Creativity (via the Art Director’s Club) has initiated to illuminate the ways in which AI is transforming the industry. I especially appreciate Emmy-award-winning director Lauren Indovina’s perspective on the evolution of tools enabling her creative expression.
“I can’t help but spread the contagion of positivity because how lucky are we?”
Indeed, where social unlocked media, and empowered the individual voice, it had optionality. It wasn’t entirely necessary.
AI feels necessary.
Our work engaging in and training it is necessary.
We’re in a phase of reinvention, as Paul Roetzer from the Marketing AI Institute puts it. We are on the cusp of reinventing knowledge work. We’re wading into realms of reinventing what it means to create (as we’ve done before with smartphones, computers, cameras).
I mean, who among us wouldn’t prefer to offload annoying work so we can focus on more rewarding, more inspiring, more useful outcomes? The wise
offered a handy primer for “training your AI intern” a few weeks back. It’s an excellent place to get started. He reminds us, “just like real people, your AI intern is not an infallible machine, but a weirdly flawed entity. It will have strengths and weaknesses you need to discover, and you will also need to teach and interact with it.”
Earlier this month Ipsos released a survey noting a majority of Americans are concerned about AI’s impacts, but only 16% of Americans had ever used a generative tool like ChatGPT, Midjourney, et al. Then last week Checkr announced, “85% of American workers have used AI tools to perform tasks at work. Millennials lead the group with 89% saying they’ve used AI at work.”
I’m tempted to ask Bing to compare survey methodologies. 🙄 Regardless, what I’m taking away is a keener interest in engaging. This is good.
Speaking of, kudos to Wired for its thorough policy describing how staff will and won’t leverage AI-generative technologies. If you’re looking for a policy, this is a smart place to start.
Photoshop Generative AI: Your results may vary
My TikTok feed was suddenly filled with photo editors, graphic designers and Adobe Photoshop enthusiasts celebrating the program’s new Beta “Generative Fill” capability. It looks impressive! (And requires downloading the Beta version of Photoshop.)
My take away—prompts matter. I’m reminded of the old Lee Stokes Retouching ad from agency Hunt Adkins, which featured a photo negative of a gorgeously lit building brick. And the accompanying typewritten letter: “Dear Lee, it seems we took a picture of this brick when, in fact, the client wanted Cindy Crawford. Can you help?” You might struggle to generate some things, but you’ll be pleasantly surprised with what’s possible now.
Think about the existing photo libraries owned by a corporation. Now all of the (formerly) vertical shots could evolve to accommodate horizontal scenarios. And vice versa. The constraints are vanishing.
🤖 vs 😬 Ad Turing Test Update
A few weeks ago I mentioned the BrXnd AI event Ad Turing Test.
Now it’s your turn to guess who made the ads: AI or Human?
Go take the test before you read further.
⏲️
Here’s how I did… Slightly better than the jury and the public who averaged 53% accuracy.
I thought humans made the first ad—mostly because of how the type was laid out. Turns out it was entirely generated by AI, and here’s how they did it. I also thought ads 3 and 4 were AI-generative because of the image aesthetics… and it turns out, well, humans created the layouts (likely using AI tools).
I suspect the humans, and the AI, are only going to get better.