The AI Revolution Might Be a Mess
Why AI safety is broken, faith in large models is waning, and what it means to be “sustainably relevant.”
Dear Friend,
The other day, Jeffrey asked me what I see as the “hot topic” with our clients these days. Yes, of course, everyone is talking about AI and the state of the world at large. But the really interesting question being brought up again and again is how to maintain what we like to call “sustainable relevance.” The question of staying relevant in an increasingly uncertain and fast-moving world, in a way that doesn’t burn you and your organization out. It is something we have been thinking about for quite a while. Of course, there is much to unpack here, but at a high level, it seems to me that the best way to navigate these tumultuous times with grace is to set the course, empower your people to find and capture local maxima, and stay in constant flow. As Bruce Lee observed:
“Empty your mind. Be formless, shapeless, like water. If you put water into a cup, it becomes the cup. You put water into a bottle, it becomes the bottle. You put it in a teapot, it becomes the teapot. Now, water can flow, or it can crash. Be water, my friend.”
And now, this…
Headlines from the Future
AI Safety Is a Mess: The uber-popular workplace tool Notion rolled out an ambitious AI upgrade with its recent release of Notion 3.0. The software offers a set of AI agents to work alongside you, the user. Sounds great (and like something from the future) – and it also opens the door to some rather nasty security issues.
A security researcher successfully got the Notion LLM agent to access private (and confidential) data and then shared this data with an external website – all through some clever, but ultimately not complicated, prompt injection.
This research exposes a fundamental security gap in AI agent architectures where traditional access controls become ineffective once agents gain autonomous tool usage capabilities. The combination of broad permissions, tool access, and susceptibility to prompt injection creates a “perfect storm” for data exfiltration attacks. Fun times. Maybe think twice before letting AI agents run wild with your data.
↗ The Hidden Risk in Notion 3.0 AI Agents: Web Search Tool Abuse for Data Exfiltration
—//—
Small is Beautiful: On the topic of (large) LLMs – a trend which is growing quite substantially is to go from large to small language models. Not only do they run on much cheaper hardware, but they can also be tailored more easily to an organization’s specific data and needs, and they also produce less surface area for vulnerabilities.
The slowing pace of improvement at the bleeding edge of generative AI is one sign that LLMs are not living up to their hype. Arguably a more important indication is the rise of smaller, nimbler alternatives, which are finding favour in the corporate world. […] As David Cox, head of research on AI models at IBM, a tech company, puts it: “Your HR chatbot doesn’t need to know advanced physics.”
↗ Faith in God-like large language models is waning
—//—
The AI Coding Revolution Isn’t Quite Happening: One of the common stories you hear about AI is that it uplevels particularly junior folks – lifting them to the level of much more experienced people (for example, Stanford found this to be true in call centers). Sounds good and logical – and yet, it might not be the full picture. Here is the counter-argument (in this case for developers): “[…] instead of democratizing coding, AI right now has mostly concentrated power in the hands of experts.”
↗ AI Was Supposed to Help Juniors Shine. Why Does It Mostly Make Seniors Stronger?
—//—
We Start to Drown in Workslop: First, we had spam (human-generated garbage content), then AI slop (AI-generated garbage content – much easier and cheaper to produce), now we have Workslop (all the crap your colleagues produce using AI). And it is costing us dearly:
Research from BetterUp Labs and Stanford found that 41% of workers have encountered such AI-generated output, costing nearly two hours of rework per instance and creating downstream productivity, trust, and collaboration issues.
↗ AI-Generated “Workslop” Is Destroying Productivity
—//—
ChatGPT Destroys Your Marriage: Yep, sorry, clickbait. But this exchange, in a recent article on Futurism, reminds us just too much of the 2023 South Park episode “Deep Learning” where Stan begins using an AI chatbot to write text messages to his girlfriend, Wendy, after seeing that Clyde is using one for his girlfriend, Bebe. Predictably, hilarity ensues. Life imitates art:
A husband and wife, together nearly 15 years, had reached a breaking point. And in the middle of their latest fight, they received a heartbreaking text. “Our son heard us arguing,” the husband told Futurism. “He’s 10, and he sent us a message from his phone saying, ‘please don’t get a divorce.’” What his wife did next, the man told us, unsettled him. “She took his message, and asked ChatGPT to respond,” he recounted. “This was her immediate reaction to our 10-year-old being concerned about us in that moment.”
↗ ChatGPT Is Blowing Up Marriages as Spouses Use AI to Attack Their Partners
—//—
New Substack Alert: Our friend, colleague, collaborator, and acclaimed Stanford d.school teacher Lisa Kay Solomon launched her Substack “How We Future” – knowing Lisa and her passionate, optimistic, and empowered perspective on the future, combined with her deep insights into the tools and frameworks which make the future not an abstract but tangible and achievable outcome, this is one you don’t want to miss.
—//—
Build Your Own Trend Radar: Wire up some AI agents, and you too can be the proud owner of your very own trend radar. Nifty!
↗ How we Built our Own Technology Radar
What We Are Reading
🔄 No Lectures, Exams, Essays: Inside a Twenty-First-Century University The anti-university: How learning by doing is revolutionizing higher education. @Jane
🤼 The Rush to Return to the Office Is Stalling Big companies are tightening return-to-office rules, but workers are still largely resisting – revealing a high-stakes standoff over the future of work that’s reshaping corporate culture and careers. @Mafe
🛢️ The World’s Appetite for Oil May Be About to Shrink China’s emergence as an electrostate might mean the world has finally reached the year of peak oil – decades after a different path to peak oil was first predicted but with similarly sweeping implications. @Jeffrey
🚀 See the World as a Startup Does Big companies could learn a lot from startups: stay scrappy, experiment more, and actually notice the customer needs that everyone else is missing. That’s how you protect today and invent tomorrow. @Kacee
🤳🏼 From Burner Phones to Decks of Cards: NYC Teens Are Adjusting to the Smartphone Ban A fascinating insight into how teens are coping with their mobile phones being taken away from them. Creativity prevails. @Pascal
Down the Rabbit Hole
👨🏼💻 Jeff Raskin, famed creator of the Macintosh (at least the initial version of it – until Steve Jobs kicked him out and took over), created some truly inspired ideas for user interfaces of the future. Now you can try his ideas in your browser.
⌨️ Talking about Raskin, did you know about the “great war for your thumb”? Steve Jobs, Jef Raskin, and the first great war for your thumbs
📺 Learn by watching carefully curated YouTube videos without any distractions: FocusStream.
🍻 It’s not just us, who like the occasional drink: Chimpanzees may consume equivalent of 1 alcoholic drink a day from fermented fruit, study finds
🍁 If you enjoy the fall foliage and are in the US, here is the Fall Foliage Map 2025 (updated daily)
🦉 Edward Packard’s “Nine Things I Learned in Ninety Years”
⛪ No Pope AI for you: Pope Leo refuses to authorise an AI Pope and declares the technology “an empty, cold shell that will do great damage to what humanity is about”
💾 Admittedly pretty nerdy, but fascinating nonetheless: How AWS S3 serves 1 petabyte per second on top of slow HDDs
🤮 This: Everything that’s wrong with Google Search in one image
📜 Ever wondered how the (dying) art of handwriting evolved? 800 Years of English Handwriting
🛰️ Amazing visualization of everything that is going on in low Earth orbit: Low Earth Orbit Visualization
🔬 Tread carefully – but it’s still exciting: HSBC claims quantum trading breakthrough