158: Four AI culture questions for leaders
Productivity, Psychology, Knowledge, Customers
We talk about culture eating strategy for breakfast, lunch and dinner. I think this has become especially true with the advent of AI. Just not evenly distributed yet, as the saying goes. But the culture-breaking pattern feels evident; perhaps because we’ve seen it before.
Once again, so many of our ingrained, preconceived, habitual bureaucratic ruts are starting to appear quaint in the face of technologies which significantly alter culture. The difference this time is at least twofold: 1) the pervasiveness of AI; it fits into almost every role, in any context; and 2) the speed with which AI has taken root.
Yesterday I had the honor of leading a conversation about AI with a mix of corporate leaders based on what I suspect matters most: Culture. As part of my work, I prepared four questions. Maybe they’ll help you and your team work through this time and place we find ourselves within. (I’ve reorganized their sequence here based on what I learned from yesterday’s conversation.)
1. What old tasks could we eliminate, and more importantly — what new tasks could we evoke?
This is the obvious productivity question. What happens if we stop summarizing meetings? Drafting formulaic documents? Organizing research data? If Copilot can infer our intent from a Teams transcript and create a Powerpoint deck based on files we’ve referenced, using (brand) style we’ve articulated, and images we typically use—what might we accomplish with the time saved? If Google’s NotebookLM podcasts can output a more inspiring creative brief from the consumer research and focus group data—what untapped value could be harvested by agency planners? What if people who previously said, “oh, I’m not creative” could express what’s in their head with greater clarity—and gain alignment faster?
I’m reminded of 15% time. Both 3M and Google used to tout the benefits of (mostly) scientific employees using 15% of their time on unrestricted creativity. What if, because of AI, every employee had 15% time?
2. How would an AI organize your personalities to unlock greater potential?
So often, established organizations are caught off guard when a competitor realizes it can function, compete, and win without a process, or roles, or hierarchy your organization assumed were carved in stone. It’s important to remember AI isn’t invested in your longstanding traditions. It has no loyalty to your org chart. Use this perspective to your advantage.
There’s a popular thread going around this week suggesting you prompt ChatGPT:
“Based on our previous interactions, what do you know about me that I might not know about myself?”
If you’ve been interacting with ChatGPT with any regularity, you’ll likely get some pleasant ego-stroking in return. You might also learn something.
But you could use a more nuanced approach. For example, take NotebookLM and feed it anonymized performance reviews, meeting transcripts, and examples of each individual’s written work. Then maybe feed it a proxy for constructs like Predictive Index, Myers-Briggs, DiSC or Caliper.
Then—ask your AI to structure a team, department, or the organization to maximize potential, or defeat competition, or whatever. What have you got to lose by trying?
3. What does it mean for your firm when knowledge becomes a zero cost commodity?
Here’s an edit of a video from
Jones which instigated this question. (And here’s the full source of his proposition—The New Tech Job Guide: How to Win in an AI World. You should watch the whole thing. It’s 10 minutes of very smart observations.)This question bothers people the most.
And for good reasons.
What does it mean when entry level talent can leverage AI to immediately understand and employ your firm’s concepts, its models, its frameworks, history, can almost immediately understand your competition, and can use AI to not just comprehend but create useful strategies that previously required years of knowledge acquisition to deliver?
And I think it boils down to two components: 1) There’s the raw material—the dates, ingredients, numbers, facts, quotes. Clearly that’s a matter of data organization, access, summarization, and display. Tasks AI is absurdly good at. Hard to begrudge anyone getting up to speed faster. 2) But then there’s taste, acumen, erudition—the ability to distill what really matters, and why, in this moment. The question becomes: Can an AI really do that; or enable employees that deeply?
We are finding out.
Remember “let me Google that for you?” The bar of expectations got raised a little higher—and even higher still once that technology was in everyone’s back pocket. What if the new bar isn’t what you trained to know, but how you solve the problems. What if certifications and pedigree are supplanted by demonstrations of capability?
4. How will our customers use AI to change us?
I wonder if this might be the most important question.
Because everyone’s customers are already leveraging AI.
Think about how the Internet changed the nature of and the definition of a customer. And then the smartphone and social arrived, further strengthening and redefining their role.
Your customers have the same AI you have—the same AI as congress people, the same AI as the CEO of Goldman Sachs (hat tip to Ethan Mollick). What changes when your organization finds itself collaborating with AI-strengthened customers?
AI+Creativity Update
🧐🤖 “The illusion of understanding” via ArsTechnica illuminates recent research from Apple, which questions the “reasoning” capabilities of frontier LLMs. In short, it’s possible, “LLMs don't actually perform formal reasoning and instead mimic it with probabilistic pattern-matching of the closest similar data seen in their vast training sets.” And those patterns are very easy to disrupt. This research doesn’t suggest LLMs aren’t effective, but it does question recent assertions around these systems ability to “reason.”
🤖📻 “Audio Generator, eliminates creative barriers for advertisers,” says Amazon (italics mine)—by which they mean, “a lengthy and complex process that costs money and often takes weeks.” In other words, those troublesome copywriters, producers, voice talent (and their agents and bookers), audio editors (and their firms), account execs and project managers (and their agencies). “Within just minutes of submitting an ASIN, Audio generator produces ad copy and a high-quality 30-second audio ad at no additional cost.” Announcement here—which indicates audio is first, video creative soon after. All of this was inevitable, but still feels like a Kurt Vonnegut plot line.
🤖🐝 Agents are coming. OpenAI announced, “An educational framework exploring ergonomic, lightweight multi-agent orchestration.” Sounds fun! I am not a coder, but I think it’s wise for everyone to occasionally look at code to better understand what’s happening—it’s not hard to grok.
They’re asking us to imagine, “situations dealing with a large number of independent capabilities and instructions that are difficult to encode into a single prompt.” i.e.
A multi-agent setup for handling different customer service requests
A customer service bot which includes a user interface agent and a help center agent
A personal shopping agent that can help with making sales and refunding orders
🤖 I just discovered educator
’s super helpful videos about a range of academic topics through an AI lens—here are three to start:🤖😆 If you haven’t played around with Google’s NotebookLM audio generator, here’s a quick primer. This is the infamous female and male co-hosted “podcast” capability that’s honestly very useful. But, of course, we need our robots to experience existential dread because that’s funny. Here’s the source link.