|
|
March 10, 2026 |
AI and the Wizard of Oz |
|
Over a hundred of you joined us live for last week's opening session of Seeds of Flourishing led by Bhaskar Goswami! The energy in the room was palpable. The chat was exploding. By overwhelming demand, the session recording will be made available to everyone (stay tuned, the link will be posted in an upcoming newsletter). We're equally excited for what comes next! On March 19th, Mike James Ross joins us for an exploration of what's happening to the meaning of work as AI is increasingly replacing human intelligence. To prepare the ground for that conversation, in today's letter we want to share some thoughts we have on AI and human flourishing. You'll hear some of our skepticism. You'll discover what we are excited about. If you're on the frontline of mission-driven work wondering if AI could help, we'd like to talk with you (details below). The story we're being told about AI, with its dazzling promises and a future that's always just around the corner, has the structure of a fairy tale we all know. It looks like the story of the Wizard of Oz. Once upon a time, there was a magnificient Emerald City... |
The Emerald City |
|
|
Reading about AI in the everyday news feels like wandering through Oz's Emerald City. Something enormous is always just around the corner. A messianic intelligence is about to arrive, ready to solve civilization's hardest problems. Or maybe to wipe us out. The number of projected job losses, the booming announcements, the spectacle of a technology that is about to change everything, for better and for worse... Most of us feel anxious, excited and confused. First takeaway: This is a designed play. The combination of those three specific emotions is a recipe to capture people's attention. When you can make people feel simultaneously scared and excited, while remaining ambiguous about what is actually happening, you're guaranteed people will consume your content, watch your ads, keep coming back for more. Let us try to do the opposite here, in the spirit of helping emotions cool down. Let's look at what’s happening under the hood. The remarkable acceleration of AI in the past few years was primarily achieved through what engineers refer to as "scaling": building bigger and bigger large language models. However, there is consensus among experts that scaling has hit a wall. OpenAI co-founder Ilya Sutskever told Reuters: "The 2010s were the age of scaling, now we're back in the age of wonder and discovery". A.k.a.: unless for some new breakthrough in mathematics, we're stuck. The epitome of this stuck point was possibly GPT-5's hyped-up launch in August 2025, followed by widespread disappointment. As tech commentator Cal Newport puts it, it's a great piece of technology, but not the leap forward we had been promised. In 2026, progress in AI relies on painstaking post-training work. These are procedures of optimization and tuning that are both human-labor-intensive and specific to narrow domains, such as coding. While we're told a fairy tale of exponential acceleration, the facts show this thing is slowing down. And that is a huge problem for AI companies. Why? Because the money is running out. The current version of AI is a disaster in terms of business. Sequoia Capital calculated that AI companies need $600 billion in annual revenue to justify infrastructure spending. But the revenue AI companies are generating is a fraction of what they are spending, and the gap is getting worse, not better. In the first half of 2025, OpenAI generated $4.3 billion in revenue but reported a $13.5 billion net loss, with a burn rate of $13.8 million per day. All major AI companies are entangled in circular deals that make them highly vulnerable to cascading losses (hence the term "AI bubble"). It was only 2024 when Sam Altman, CEO of OpenAI, described the idea of bringing ads to ChatGPT as "deeply unsettling" and a "last resort", recognizing how it would not be aligned with the user's best interest. Fast forward to February 2026, ads are now part of ChatGPT. The Emerald Palace is crumbling. The only way to keep this going is to convince us that we should really want this. And for that, you need a great story. That's where "The Thinking Game" comes in. |
|
|
Freely available on YouTube, this documentary is a high-end example of the attempt to touch that ultimate emotional chord. The film traces Google DeepMind's founder Demis Hassabis from child chess prodigy to Nobel Prize co-winner for solving the protein folding problem with AlphaFold. Demis' story is inspiring and worth watching. AlphaFold is a real breakthrough in biology. And yet, we found this and other similar messages troubling. Watch closely what the film does. It takes you on an powerful and relatable rag-to-riches story. It blows your mind with the marvel of AlphaFold, which is a specific, hyper-narrow application of AI...and then covertly transfers that glow over to the message: "We're building Artificial General Intelligence, it's the coolest thing ever". Wooow! But wait a second...what?! Artificial General Intelligence (AGI) is a hypothetical machine that can think, reason, and learn across any domain the same way a human can. The problem: AGI is categorically different from AlphaFold, trying to build it is costing an atrocious amount of money, we are nowhere close to achieving it, and it's far from clear if we should even want it. If anything, AlphaFold is evidence that the most powerful applications of AI are deeply specialized. It is an argument against pouring trillions into the fantasy of generalized artificial intelligence, not for it. The Emerald City is dazzling. But it has always depended on nobody looking too closely at how the light is made. And that's not all. Wait until you discover who is behind all of this... |
The Man Behind the Curtain |
|
|
There is a striking passage, early in the same documentary, where Hassabis mentions their first investor. Pay attention to the language and the vibe it creates: "We needed investors who aren't necessarily going to invest because they think it's the best investment decision. They're probably going to invest because they just think it's really cool. He's the Silicon Valley version of the man behind the curtain in The Wizard of Oz. He had a lot to do with giving you PayPal, Facebook, YouTube, and Yelp. If everyone sees X, Peter Thiel suspects that the opposite of X is quite possibly true. So Peter Thiel was our first big investor." How exciting. How reassuringly benign. It's the kind of against-all-odds story we all want to be part of, right? Maybe not. The metaphor risks being more accurate than Hassabis intended. For those who remember the story, the Wizard of Oz is an ordinary man using smoke and projections to maintain an illusion of great and terrible power, who hides behind a curtain and sends others on dangerous missions while staying safely out of sight. What the documentary fails to mention is that Peter Thiel is the co-founder of Palantir Technologies, a firm backed by the CIA's venture arm whose website describes its targeting capability as an "AI-powered kill chain". The UN Special Rapporteur on human rights has concluded Palantir technologies were most likely used in unlawful attacks causing mass civilian casualties in Gaza. Furthermore, Palantir won a $30 million contract to build "ImmigrationOS" for ICE, a system for near real-time deportation tracking that has been used in the recent controversial "crackdown" on immigration in the United States. To be fair to Demis Hassabis, DeepMind has maintained an ethics board and has been publicly vocal about AI safety. There is no public evidence that DeepMind's specific technology has been deployed in military-adjacent contexts. Whether Google, which owns DeepMind, will respect that boundary, remains to be seen. It is theoretically possible that Peter Thiel may have invested in DeepMind simply because it was "really cool". Either way, this narrative fails to mention that the charming Silicon Valley contarian in this origin story is simultaneously building the infrastructure of AI-powered mass surveillance and warfare. The story we are being told about AI has holes in the middle. Just two weeks ago, Anthropic, the company behind the chatbot Claude, showed to the world what happens to those who actually uphold the concerns for the use of AI. Anthropic refused a Pentagon ultimatum to remove restrictions that had been written into their existing contract, restrictions prohibiting the use of Claude for mass domestic surveillance and fully autonomous weapons. As a result, they were publicly insulted by the President of the United States as "woke" and "nut jobs," and designated by the Department of War as a "supply chain risk", which multiple observers regard as an unprecedented and punitive measure against a company that is simply demanding that its terms of use be respected. As you follow the money trail from the dream of curing diseases to the machinery of war and mass surveillance, the deception shatters and leaves a really bad taste. |
The Ruby Slippers |
|
|
And yet, we do see areas of exciting potential. There might be a pair of magical ruby slippers that the Dorothy of our story could activate. We see something important happening, not in the realm of AGI fantasies, but on the ground, in the places closest to the community work we care about. The pattern is clear: specialized AI-powered applications can handle increasingly complex levels of repetitive and menial labor. This means that energy could be redirected to the most vital areas of work, where resources are limited and only humans can deliver on the mission. There are numerous examples in the non-profit world:
None of this required artificial general intelligence. It required specific, useful tools in the hands of people who already had the heart, the brains, and the courage for the work. This matters because many of you are doing exactly this kind of work. You are community organizers, nonprofit leaders, coaches, educators, counselors and more. What's exciting is that the technical and financial barrier for building such tools has substantially fallen. At the Centre for Human Flourishing, we've been exploring what it would look like to help non-technical professionals learn how to build with AI. True to our identity, we're envisioning a community-driven, high-touch co-learning experience. We're in the early stages, and we're looking for partners. If you are a mission-driven organization who would love to learn how to build with AI reply to this email and we'll share more. |
March 19: Meaning at work in the age of AI |
|
|
The emerald city, the man behind the curtain, and the ruby slippers. These are all facets of the same question: What does it mean to do meaningful work in the age of AI? That's what we'll be exploring on March 19th. Mike James Ross, leadership advisor, former CHRO at Simons, and co-author of "Intention: The Surprising Psychology of High Performers", argues that our understanding of "meaningful work" is more fragile than we realize, and that AI is dramatically exposing that fragility. But rather than despair, Mike sees an opening: by returning to an older, deeper sense of purpose, AI could become a catalyst for reclaiming what makes work profoundly human. If you're not registered already, here is the link: |
|
The session is online, 45 minutes, free, and open to all, with the possibility of making a donation. With warmth, The Centre for Human Flourishing Team |
|