arrow-right cart chevron-down chevron-left chevron-right chevron-up close menu minus play plus search share user email pinterest facebook instagram snapchat tumblr twitter vimeo youtube subscribe dogecoin dwolla forbrugsforeningen litecoin amazon_payments american_express bitcoin cirrus discover fancy interac jcb master paypal stripe visa diners_club dankort maestro trash

The Newsletter | Edition 092
Progress Report is dedicated to providing inspiration for action. In our Off-White Papers, we provide practical guidance on how to respond to our rapidly-changing world. This newsletter explores those topics in real-time, with information and action steps on how to make progress now.

But in this special newsletter series, The State Of, we dive a little deeper into the long-term work that comes after, in the places where we’re seeing new types of progress in action. From brand strategy to design, internet trends to sustainability, music to science, beauty to travel, and more.

I’m not sure if you’ve heard, it hasn’t really been in the news, but there have been some recent advances in the field of generative AI. It would be easy to miss if you don’t watch the news, use social media, read any articles, listen to any podcasts, or walk into any coffee shop with laptop workers anywhere in the world. To that one person who fits these criteria, but somehow also reads this newsletter, teach us your ways.

For everyone else, the past six months — ChatGPT launched to the public in November — have been a period marked by deep reflection (by some) and existential anxieties. Somehow rising in pitch and intensity, day after day, week after week.

The future has become dramatically obscured by the complexity we’ve introduced into today. Some systems and structures will fall. New ones will rise rapidly from the ashes, like a reborn Chicago after the Great Fire.

What seems unavoidable, the one thing technocrats and policy wonks alike seem to agree on, is that the game has changed. The toothpaste has erupted from the tube, as if deployed with a sledgehammer, and it’s not going back in.

If there is no turning back, is there a positive picture looking forward? Can I interest you in becoming a centaur?
After IBM’s Deep Blue finally defeated chess Grandmaster Gary Kasparov in May 1997, Kasparov said humans were “doomed.” But rather than accept his doom, Kasparov turned to folk wisdom: “if you can’t beat ‘em, join ‘em.” In June of 1998, Kasparov played in what is recognized as the first Advanced Chess game — also known as Centaur Chess — where he and his opponent each had an AI assistant. By 2005, Freestyle Chess was born, allowing greater combinations of players on teams, where teams could be composed of many humans, many supercomputers, or centaur combinations of the two working collaboratively. The centaurs won.

A lot has changed since 2005 — though maybe our pant styles are coming back? — but one thing that seems to continue to be true is that centaurs win. You can look at finance firms deploying algorithmic centaurs like at Ray Dalio’s Bridgewater Associates. Social media companies implementing advanced algorithms. Medical professionals analyzing X-ray, CT, and MRI scans with help from AI. Or simply using a calculator or your GPS to get you where you’re going. We’re constantly amplifying our human abilities with computational power.

Today, we’re sitting in that moment that Kasparov found himself in in May 1997— we’re recognizing that generative AI has surpassed us in some way. We’re seeing a bit of ourselves in it. And maybe we’re jealous, or threatened, or self-conscious. Whether you’re an artist looking down the barrel of Midjourney, a lawyer watching GPT-4 beat your LSAT score, or a consultant wondering if your strategies are on the table, the acceleration of these technologies seemingly threatens our livelihoods.

But what happens when you, as an individual, or we, as communities and teams, take the Kasparov view. What happens when you see the game as non-zero sum, and choose cooperation? A world of centaurs, composed of computational might and human ingenuity.

Many of the fears we discuss are based in zero-sum thinking: “The AI will take all of our jobs,” or “With AI, we won’t be needed any longer.” This is sometimes described as the Lump Labor Fallacy. As the St. Louis Fed describes the idea:

"The lump of labor fallacy is the assumption that there is a fixed amount of work to be done. If this were true, new jobs could not be generated, just redistributed. Those who believe the fallacy have often felt threatened by new technology or the entrance of new people into the labor force."
The idea assumes that if you can produce X-value, and an AI can also produce X-value more cheaply, a company will replace you with an AI agent. What it fails to consider is “what if a human and AI centaur can produce 5X value when working together?”

What it also fails to consider is, if the AI can tackle the parts of work that are joyless, what more could you do if those tasks were removed from your plate?

In his 2001 book Good to Great Jim Collins introduced the idea of Hedgehog companies. The concept is based on three intersecting circles (that look like a hedgehog, how cute), asking three questions:

  1. What are you deeply passionate about?
  2. What can you be the best in the world at?
  3. What best drives your economic or resource engine?

While I won’t claim to have read Jim’s book, when I recently encountered this idea, I was filled with optimism. The optimism of the procrastinator, the hacker, the short-cutter. The idea that I might not be jettisoned from the world of work, but that I might rebalance the ratio of my work, with more time spent doing the things that I love and less of the things I don’t.

That is the promise of the centaur. To build a “stack” of AI tools that support your own personal workflow, so that you can focus on those hedgehog questions and offload the things that don’t bring you joy to the AI half of your centaur. Like a digital Marie Kondo, asking whether each task brings you joy, and nicely folding the ones that don’t into a box labeled “For GPT/Bard/Bing.”

And so the question becomes, how do we all become magnificently splendid mystical creatures, all flowing hair and thrumming sinew (except, you know, with AI)?

Back in early March, one of Meta’s most powerful LLMs (large language model) called LLaMA leaked online. The result was a sudden and radical infusion of potential into the open source community. In April, an internal memo at Google began circulating. Written by a senior engineer named Luke Sernau, the memo warned not of OpenAI, Microsoft, or other expected competitors — it warned that Google and OpenAI were in threat of being left behind by the open source community that has been rapidly innovating on the LLaMA model. Why? Because this community has already created a huge volume of new tools that have the potential to accelerate AI’s distribution into all aspects of our work and everyday lives. In other words, they’re creating the toolkit for personalizing your own centaur.

My personal centaur today is based using a mix of small list AI tools — ChatGPT Plus, Bing Chat, Bard, Midjourney, and a few Chrome plug-ins. I’ve been able to run a factor analysis on a small data set with ChatGPT. I’ve been able to add a bit of whimsy into projects with Midjourney, as well as give a huge boost to a personal Dungeons and Dragons game with friends. With tools in my emerging centaur, I’ve been able to check for bias in writing, create sparks of inspiration, and accelerate analog production for creative work. I’ve been able to program complex equations in Excel that far exceed my personal knowledge base of Excel programming when working on financial models. And I know I’m barely scratching the surface.


  1. Play with some tools. I find that using the tools for creative side projects has accelerated my understanding. I’ve written silly songs about friends with GPT. I’ve created AI art for games and input funny conversations or turns of phrase into Midjourney. Play is, afterall, just a long series of experiments strung together. Once you get a handle on what can be done, it becomes a lot easier to see the utility in your workflow vs. starting workflow first.
  2. Become fluent in prompt engineering. Or, more simply put, how you design your requests into these different models. At a fundamental level, this is collaboration. The models will do exactly what you tell them, but ONLY exactly what you tell them. This means your creative interrogation needs to be precise in a way we often aren’t with each other.
  3. Don’t expect perfection. All of this is great at creating sparks, accelerating my own creativity, doing rote tasks that I don’t like (editing), or specialized tasks I’m inexpert at (Excel programming). But today you still need to be concerned with hallucinations (e.g. false information), and in the creative field you can’t anticipate deeply moving or strategic caliber output. However, it’s an accelerated start to tasks. As a chronic procrastinator, what more could you ask for?!

While optimism might be a bold stance in a sea of uncertainty, it doesn’t reduce the need for a deliberative, cautionary approach. What I think it can do is allow us not to act out of fear and find opportunities for non-zero sum, mutually beneficial solutions. The centaur, if it really is the way our evolution progresses in some trans-humanist dream, is a way to find and amplify the best of what we as humans bring to the table. I don’t know, I like that future better than the alternatives, so I’ll be rolling with it for a minute.

Shopping Cart