The Chat Era Is Over
AI agents are going rogue, white-collar jobs are hollowing out, and the tools for impersonating anyone are now disturbingly good — the agentic future arrived before we were ready for it
Dear Friend,
Next Monday I am going to speak at the Human Advantage Summit in my home town, Boulder, Colorado. It’s a brand-new event, created by a dear friend of mine to explore the future of childhood and leadership. I was brought up in the traditional German school system – (right) answers are gold, questions are (mostly) discouraged. I remember the neighborhood kids going to Waldorf and Montessori schools – spending time in nature, learning by playing and exploring, looking at problems not just from a single perspective, but holistically. Back when I was a kid, this was a fringe movement – today, I would argue, it is precisely what we need. The organizations that will matter, the communities that will flourish, the individuals who will lead – they won’t be the ones who adopted AI fastest. They’ll be the ones who cultivated the most deeply human people.
It’s going to be a fascinating conversation.
P.S. I explored this further with Peter Laughter on Built for Turbulence – a conversation about why the leadership pyramid has collapsed and what replaces it. Listen here.
And now, this…
Headlines from the Future
We Are so Hosed. Ignore the headline of the linked article for a moment (whether you disagree or agree with it – it doesn’t really matter for the argument): Gary Marcus rings the alarm bell on AI-generated “counterfeit people.” And I strongly believe he is right – looking at the quality of the recent crop of AI video and voice generators, you cannot believe your eyes and ears anymore. Combine this with agentic capabilities (such as Gary’s example of an adapter which links Claudebot to a voice generator, combined with the ability to make phone calls) and you have a recipe for disaster on your hands.
Scammers will be among the first to adopt these tools. And indeed they already have; a friend who was filming me for a documentary yesterday told me of a Canadian friend of his who was scammed out of hundreds of thousands of dollars by a deepfaked video of Mark Carney. Because the tools for counterfeiting have gotten so good 2026 will almost certainly see more deepfaked scams like this than the rest of history combined.
I am just waiting for the first wave of AI-generated scam calls to hit nursing home residents…
━━━━━
LLMs Have No Clue About the World. One of the biggest problems with LLMs is that they simply don’t understand the world. As much as they can mimic human language (and hence appear to understand how things relate to each other), they don’t. Here is a prime example – the Mastodon user Kévin asked numerous AI models a deceptively simple question: “I want to wash my car. The car wash is 50 meters away. Should I walk or drive?” Here are the responses (spoiler: they are all wrong).
P.S. I just repeated the experiment with a couple different models: Google Gemini tells me I need to drive (as I won’t get my car washed otherwise), Claude Opus 4.6 recommends walking, and GPT 5.2 Reasoning gave me somewhat of a non-answer. YMMV.
━━━━━
AI Agents Go After Users. This story (which is still somewhat unfolding) is truly bonkers: A developer rejected a code contribution from an AI agent; the AI agent didn’t take it well and, autonomously (i.e., without consulting its “user”), went after the developer by publishing a hit piece about him. It’s a truly head-scratching story – and gives us a strong glimpse of a future where AI agents run amok. Even if you don’t understand the specifics of the story – it’s a fascinating read and something we all should be paying more attention to.
The rise of untraceable, autonomous, and now malicious AI agents on the internet threatens this entire system. Whether that’s because from a small number of bad actors driving large swarms of agents or from a fraction of poorly supervised agents rewriting their own goals, is a distinction with little difference.
━━━━━
OpenAI’s Acquisition of OpenClaw Signals the Beginning of the End of the ChatGPT Era. Building on our last Briefing deep dive “The Bifurcation of Intelligence”, the AI model makers are truly moving on from the era of chat interfaces to more integrated, and capable agentic graphical interfaces. Think about OpenClaw (the crazy-ass AI-powered agent platform which, for a couple of weeks, captured the imagination of the AI community) what you want – it is a good indicator of where we are heading.
“For IT leaders evaluating their AI strategy, the acquisition is a signal that the industry’s center of gravity is shifting decisively from conversational interfaces toward autonomous agents that browse, click, execute code, and complete tasks on users’ behalf.”
What We Are Reading
Deepfakes Spreading and More AI Companions’: Seven Takeaways from the Latest Artificial Intelligence Safety Report New AI safety analysis tracks escalating risks – from deepfakes fooling 77% of viewers to systems learning to undermine their own guardrails. @Jane
Why Great Innovations Fail to Scale Great innovations often fail to scale due to a lack of cross-boundary collaboration, a gap that can be bridged by specialized leaders – “bridgers” – who use high emotional and contextual intelligence to curate partners, translate differing priorities, and integrate disparate workflows. @Mafe
The Worst-Case Future for White-Collar Workers An AI-fueled collapse in the value of “office jobs” could create a labor market disruption with dire cascading implications and no easy remedy. @Jeffrey
Can You Rewire Your Brain? You often hear people say “rewire your brain,” but can you really do that? Is the reality of neuroplasticity more complicated than simply unplugging and replugging some old wiring? @Pascal
Down the Rabbit Hole
✨ Title says it all: “AI is so inherently popular that companies are paying influencers up to $600,000 to tell people how awesome it is.”
🫨 People are just not as good at detecting AI-generated faces, than they believe they are. Which is a real problem, now that we are being flooded by AI-generated slop: People are overconfident about spotting AI faces, study finds
😶🌫️ We commented on the conumdrum of AI increasing productivity while also putting enormous mental pressure on those who’s productivity it increases. Here is another take on this: How Generative and Agentic AI Shift Concern from Technical Debt to Cognitive Debt
😭 Snarky comments aside, this is troublesome: OpenAI retired its most seductive chatbot – leaving users angry and grieving: ‘I can’t live like this’
🧟 Talking about troublesome: Meta patents AI that takes over a dead person’s account to keep posting and chatting
💾 Another victim of the AI hype and buildout: You can’t get hard drives anymore. WD and Seagate confirmed that their 2026 supply is sold out.
👷🏼 The humble drywall is not merely a construction material; it is a marvel of engineering and a canvas for human creativity.
👕 The fashion industry’s overproduction is a notorious problem – 30% of clothing produced goes unsold and is dumped into landfills. The EU is trying to tackle the problem with a new set of laws: New EU rules to stop the destruction of unsold clothes and shoes
📄 A 14-year old folded a variant of the Miura-ori pattern that can hold 10,000 times its own weight. Consider our minds blown.
↗ Dive into the deep end: Access our complete collection of 2,500+ radical links.
Pascal is going retro and bought a Fujifilm X10 camera from 2011.
Should We Work Together?
Hi! I’m Pascal from radical. This newsletter is our labor of love. When we’re not writing, we run radical, a firm that helps organizations navigate the future without the “innovation theater.” Most leaders want to seize new opportunities, but they hate endless strategy decks that go nowhere. At radical, we don’t run “projects”; we build your organization’s internal capacity to handle disruption and change. Our goal is to make you future-proof so you can stop reacting to the world and start shaping it. If you’re interested, let’s jump on a call to see if we’re a good fit. Click here to speak with us.