A New AI Organization Beckons
A New AI Organization Beckons
AI will become your company's most valuable administrator, analyst and creative collaborator. Put it at the center of how human teams collaborate.
Welcome to People vs Algorithms #65.
I look for patterns in media, business and culture. My POV is informed by 30 years of leadership in media and advertising businesses.
Sometimes it’s nice to read in the browser.
It's been about six months since ChatGPT was released as a consumer product. A lot has happened in a frighteningly short span of time. A hundred million people have used the product, it has made its way on to every software product roadmap, reignited the search wars and destabilized the most dominant company of the modern internet era, Google, unleashed a fury of investment, opened up a lively dialog on the future of work and left pretty much every company grappling with how it will alter the fundamentals of their business.
Broadly, it’s created fundamental questions on both sides of the media business — both how AI reshapes how media is made and how audiences find and consume it. Of immediate importance is what happens to so many media businesses dependent on the movement of attention from search query to a monetizable results page. But what will become more important in the months ahead is how leaders quickly restructure people and systems around AI inside of their organizations.
This week a friend asked how things will look in three, five, ten years from now. If we are to extend the pace of the past six months, even three years is hard to discern. I thought it would be illustrative to put the current pace of change in context. The rise of the internet changed how we live and work pretty materially, but that evolution took twenty years.
The internet was born in 1991. Penetration reached 10% by 1995, 30% by 1998, 50% by 2001 and roughly 80% 20 years later in 2010. Through the 90's and early 2000s, a slew of multi-billion dollar startups rose to prominence — Yahoo, eBay, Amazon and Netscape. Google followed in 1998, LinkedIn in 2002, Facebook in 2004, YouTube in 2005, Twitter and Spotify in 2006, the iPhone hit in 2007, Airbnb in 2008, Uber in 2009, Instagram in 2010. The amount of new value created was staggering over that 20 years. The way we thought about media, commerce, mobility, marketplaces, work and trade changed pretty fundamentally. But it took a while. Because all the rails were new. This change is gliding on top of them.
You are probably not alone wondering how best to factor this change into your org. There's the obvious of 1) talk about it; 2) get everyone to use it; 3) deconstruct how it impacts development, creative, sales, and operational roles; 4) anoint enthusiasts and fund experiments across the org, etc.
But we need a new mental model for the AI enterprise, one that reflects AI’s emerging role as an infinitely flexible brain sitting at the center of the organization, the people inside of the organization as planets that move in its orbit.
This idea came to me in a discussion with friend (and podcast collaborator) Alex Schleifer. Alex mentioned how central AI was becoming to everything they were doing inside of his creative / gaming studio, Universal Entities. It’s a small and technically savvy group, accustomed to distributed work and connected with internal collaboration tools like Discord which has become the company's main communication backbone. The’ve baked the generative AI visualization tool Midjourney into their team Discord instance, and now can instantly visualize any concept as part of routine communication. The team is training ChatGPT based models to manage their game narratives and character libraries. AI will become the real time, living neural container for company IP. Anybody can get the backstory on a character by asking the chatbot. More interestingly, anyone can enter into a dialogue around how a character might behave and why, what a new character might look like, paths along which an arc might unfold, etc. This is not a necessarily deterministic process. It is a back and forth between person and machine. One with human context and history.
Here, AI is quickly becoming a central intellectual infrastructure to the business, a repository of project knowledge and more — like a natural language intranet with a huge brain that deeply connects internal knowledge work to a collective human repository of thought. It's a combination of knowledge base, helpful assistant and creative collaborator. Alex made the point that the “internal brain” would likely connect to an “external game brain” that powers player interaction post launch. This is a powerful concept. Like selectively pulling down the walls between internal knowledge systems and intelligent customer support in any organization.
Our conversation reminded me of something I built at Hearst about five years ago that I think suggests where things are going. Frustrated with an inability to get easy-to-consume analytical data in the hands of groups across the org, and the lackluster adoption of complex reporting dashboards that no one seemed to use, I decided we should build all of our data endpoints into our core internal communication platform, Slack.
Slack had become foundational to how we communicated across every internal function. It occurred to me that if we could build a semantic front end to all of our data AND pushed it into the environment where we spent the day communicating and managing the business, we would all get smarter and spend more time on the right things. We pulled everything into Slack — content performance data (volume, content category, audience data), revenue data (advertising, subscription, affiliate and e-commerce revenue) and traffic source data (search, social, direct, email) as well as operational data (what’s for lunch in the cafe) into Slack. We called the tool “HANS,” a friendly acronym for Hearst Answers.
Anyone in the organization could query "Top beauty story on Harpers Bazaar," or "How many blenders were sold on GoodHouseKeeping this week," or "How much traffic did we get from Google on Esquire today," etc. inside Slack. The entirety of our analytical repository becomes available as an answer to a question inside of our chat system. Data extended effortlessly into conversations between teams in Slack channels. Alerts could be set to deliver notifications automatically.
Read the case study the team at Slack produced on our project here.
This power of smart, natural language interfaces to all types of organizational data and business processes has become immensely more powerful with maturation of large language models and the ability to augment internal IP with a world of knowledge outside the company. The ability for machines to make sense of data has evolved 1000x since we embarked on our Slack project.
There's an interesting conceptual model here, one that I think will come to define organization structure over the next five years. The modern enterprise will be defined as a data set and system of business rules with AI at the center as orchestrator. AI will function as a dedicated and infinitely knowledgeable employee positioned alongside human collaborators inside of internal communication infrastructure like Slack, Discord or Teams. The best and most nimble companies will make generative tools part of every dimension of the organization, sometimes as human replacement but more often augmenting human creation and judgment in a way that makes people significantly more informed and productive.
Enjoy the weekend…/ Troy
ON THE PODCAST
AI meets tech's friction paradox
Nobody likes friction, at least in the moment. It stands in the way of accomplishing a task. Modern life in many ways has been an exercise in applying technology to remove friction from everyday processes. There has long been too much friction in media businesses, not just in delivery but in the service of making as much money as possible. This week, we discuss the positives and negatives of friction in media products. Listen here.
ON DECK
A one man Muzak machine
Yesterday, Mac DeMarco dropped 199 songs of unreleased material entitled “One Wayne G,” a nine hours dump of fresh sonic wallpaper.
Mac is living the "creator" dream. Making shit he owns and wants to make, experimenting with formats in public, serving loyal fans, unencumbered by record company doctrine.
Below, an PBS interview from a few years back in his LA studio.
THE USUALLY PROLIFIC Mac Demarco made fans wait an uncharacteristic four years before releasing his new album Five Easy Hot Dogs, earlier this year. On Friday, DeMarco made amends for the lengthy hiatus by data-dumping a half decade’s worth of unreleased music on streaming services.
DeMarco’s One Wayne G spans 199 songs stretching over nine hours. As an Instagram Stories post from Mac’s Record Label revealed, the compilation was named after NHL legend Wayne Gretzky, who starred for DeMarco’s hometown Edmonton Oilers in the Eighties. More >>
Source: Rolling Stone