Marc Mathieu

Control is not a forecast. It is a decision. A probability makes it sound like weather, something that happens to us. It is closer to architecture: built, or not, in thousands of choices about what a system is for and who can stop it. Weather forecasts are useful. Nobody has ever changed the weather by reading one. Which of those choices matters most right now?

Claude leads around a quarter of Anthropic's own AI R&D. Humans still decide what to ask and what to act on. But the higher that share goes, the harder those decisions get. The machine can do a good part of its own homework. Someone still has to mark it. So who decides, and who makes sure it serves workers and people at large?

The most important safety feature in AI may sit outside the model. Separate the model from its harness. Keep controls external. Log every meaningful action in human readable form. Let an authorised person pause it mid-task. An emergency brake. The human moves from inside the black box to holding the controls and the record. Every car comes with a brake. We still decided who gets to drive. So who should the authorised person be: the lab, the company deploying the agent, or the people it acts for?

The saga about the long horizon verifiability / reliability of agentic AI continues with Emergence World Season 2. Fascinating to be involved personally with a start up / a research lab at the frontier of the agentic revolution, pointing to the awes... and flows of ai inventions! Is Neuroformal AI the solution - i.e. as described below "combining the flexibility of neural systems with formal methods that can provide rigorous, observable and auditable verification as autonomous systems act"?

Today is Labor Day, won by workers whose leverage is now disappearing. Almost every AI leader spent the summer promising abundance. But do we want a future of abundance in which we have no agency?

Congrats on ClaudeForce @marc ! Great move (and I am assuming more to come!). Would love to reconnect next time you're in NY. I've been busy investing and advising Al Frontier Startups! Even doing V2MOMs with some of them 😉 So much opportunity in this age of AI.

I've been sitting with this essay since it dropped. Most of the commentary is about the framework and the AGI timeline. The framework is useful. The timeline is indicative. But the real substance is in the deeper questions, the ones too often left undiscussed: How should society adapt to a world of abundance? What values should guide the most powerful AI systems? How will work, meaning and the human experience change? Demis is right that these aren't questions for technologists alone. A shift with 10x the impact of the Industrial Revolution, at 10x the speed, will redefine human potential, human experience and human identity. These questions are the very reason I could not not start Rogue Ventures. The frameworks will sort themselves out. The human questions demand our attention now.

I read the ai-2027.com scenario, and I have to be honest, I came away disappointed. Not because the research isn't rigorous. It is. But because it continues to think of AI as a tool to do what we already do, what we already know how to do, just better, faster, cheaper. Even when it imagines AI as a system, it imagines that system completing work as we know it, flowing through workflows as we know them. It never quite reaches for something radically new and different to this world. I think there are two reasons for that. ➡️ The first is the horizon. 2027 is simply too short. What that timeline describes is not AGI as I would conceive it. At best it is what I would call AWI, Artificial Workflow Intelligence, the automation of work as it exists today. ➡️ The second is that a single transformation curve cannot hold what is actually coming. You would need three parallel curves, with three different horizons, each starting now. One for Deploy. One for Augment. One for Transform. We saw this with electricity. The factories that simply swapped steam for electric motors captured a fraction of the value. The ones that reimagined the factory itself changed the world. I truly believe the real story of AI will not be written by those who make today's work more efficient, but by those willing to imagine the work, and the world, that doesn't exist yet. What are your thoughts?

In a recent address to a group of senior business leaders, I described the world we all came from that no longer is. The consumer was consuming, so we built mass segmentation. The audience was listening, so we built mass broadcast to target those segments with highly produced advertising. The shopper was shopping, so we built mass distribution to promote to them. Digital and social commerce began to unwind that system. E-commerce went further. AI, with its capacity for hyperpersonalisation and agentic orchestration, is finishing the job. The audience is not listening anymore, and the ad is disappearing, embedded directly into content and experiences. The shopper is no longer shopping: marketing, sales and customer service have merged, and the transaction now lives inside the conversation. And the consumer is no longer consuming. Nobody wants to be ‘targeted’ (hate that word!) as a segment anymore. They want to be understood as an individual, their needs read sometimes better than they read them themselves. I spent decades in rooms where the campaign was the craft. The targeting, the buying, the optimising: that was the work. But today, the platforms are no longer selling marketers tools. They are selling us agentic workers. TikTok launched a marketplace of ready-made AI agents that build campaigns, generate creative and manage catalogues. Meta's AI now buys media autonomously, and it is on track to overtake Google in global ad revenue on the strength of it. Hand over the assets, the goal and the budget, and the machine does the rest. Watching that work migrate into a loop that runs without us made me pause, not because the machine does it badly, but because it does it well. So what remains? I keep coming back to the same answer. The machine can run the campaign. It cannot yet decide what the brand is, and who it is for. Judgement about meaning, about what a brand should be, and who it should be for, moves from one job among many to the job. The marketers who thrive in this next phase will not be the ones who operate the tools fastest. They will be the ones who know what to ask the agentic workers for.

I am an optimist about AI. That is exactly why I refuse to pretend the next few years will be painless. What worries me most is not how long people take to adapt. People adapt remarkably fast. It is how long systems take. Our economic models and our governance models will lag behind this technology, and in that gap there will be real disruption to real livelihoods. Saying so out loud is not pessimism. It is the entry fee for being taken seriously as an optimist. Because the pattern is one we have lived through before. Music was declared dead when records collapsed, and it came back transformed, bigger than ever, through streaming. I truly believe AI will, in the long run, enable humanity to thrive. But the adaptation period is coming either way. The choice we actually have is whether we prepare our systems for it, or let the gap do its damage first.