139: For the 32nd time
Travelled to MKE; Also—can AI be creative?
I was in Milwaukee yesterday, giving a generative AI talk to a room full of CFOs. I’m grateful to the Wisconsin Chamber of Commerce for the opportunity. This was my 32nd such presentation since March 2023. The quality of curiosity has shifted from “wait, what?” to “can we get specific?” En route I finished listening to episode 175: “Everyday People” by Sly and the Family Stone Part 2, “My Own Beliefs Are In My Song” from Andrew Hickey’s fabulous podcast A History of Rock Music in 500 Songs. Strong recommend.
The through line is human behavior
My presentation, Generatively Better, has evolved significantly since its inception. Only four slides remain from the first iteration. But the central theme hasn’t changed: The impact of generative AI is much less technical and much more social and cultural. The technology is actually trying very hard to become invisible. How we think and how we operate together in this new context is what truly matters.
Everyone has dabbled with AI tech. But as Benedict Evans notes, the vast majority haven’t yet found a strong reason to commit. It’s not essential…yet, for some reason. (Let’s remember it took years for the iPhone to become ubiquitous.)
Every CFO I spoke with last night acknowledged their organizations were, of course, piloting all manner of generative AI projects. Few were personally investigating AI’s potential, however. This follows every tech trend I’ve experienced—despiteChatGPT’s historic early wins, it turns out humans are generally cautious. We’ll let someone else, another team, take the perceived risks first.
AI+Creativity Update:
🤖✏️ Noah Brier (founder of Percolate, and BrXnd the AI Conference) writes persuasively on the nagging question: Can AI really be as creative as humans? Of course this is nuanced territory. As Brier puts it, “Creativity is a process—often an incredibly confusing one.” It’s folly to expect a technology to immediately encompass, and reliably deliver, what has been an achingly complex and delicate human endeavor. Yet here were are!
🤖 Shelly Palmer has curated a very useful set of prompting formulas. Each is an acronym, giving you not just a structure for writing but a broader means of thinking through a problem. This collection reminds me of an MBA playbook—i.e. The frameworks matter just as much as the solutions.
🤖😆 “Not yet.” That’s the quick answer to Google Deepmind’s question: Can Language Models Serve as Creativity Support Tools for Comedy? (Full report here, The Neuron’s recap here.) Remember—today’s AI tools are the worst versions we will experience. As humans engage, as tests like these occur, the LLMs will improve.
🤖🤔 Ethan Mollick has penned another insightful observation, with tests you can try for yourself on the evolving edge of LLMs; specifically Claude 3.5. That we can now upload something as arcane (to me at least) as a spreadsheet into a conversational experience—be able to discuss the data, and ask for transformations to occur like dashboards—only demonstrates how radically far we have come in such a short time.