Dateline: November 27 2029
We’re facing a new financial fragility, and it’s nothing to do with questionable credit deals or even the so-called AI bubble. It’s something far more opaque, insidious and systemic. It’s called the agentic cascade.
AI agents are wonderful things. They can be tasked to monitor all sorts of digital systems, from internet traffic to electrical grids to market movements, and react autonomously to signals and events. It’s what they are programmed to do. And they’re very quick and very efficient, executing their decisions long before humans have even noticed that anything’s amiss.
But that’s where the problem lies. If an automated warehouse agent detects a surge in demand — like a Black Friday promotion — and spins up additional capacity, which is interpreted by the power company agent as an anomaly, triggering demand limiting to the supercharging network, leading to robotaxi “outages”… then you have a cascade of agentic actions that could have entirely unforeseen results.
The financial system is the most vulnerable. Now that most stock markets trade 24/7 and are integrated globally, there’s no such thing as “after hours” to force everyone to take a breath. Markets have their own agents to protect them and avoid flash crashes, but they’re not in the business of protecting each other. My agent maximises trades on my behalf, and yours does the same for you. Exclusively.
Black Swan events are random, unlikely and totally unpredictable. AI agents can recognise patterns, analyse results and model outcomes. But they can’t predict how other private agents will react to new data. So, don’t let AI agents control your financial future. They’re not prepared for the agentic cascade. / First published in Mindbullets November 27 2025.
Say goodbye to Gen AI
There’s a new agent in the matrix
Dateline: June 6 2027
The tech industry was looking decidedly mundane and mature almost a decade ago. Phones were hotter than laptops and servers in the cloud were quite passé. The biggest thing was — yawn — Big Data. Then the Covid lockdowns happened, and tech got a shot in the arm.
Suddenly it was about remote working, teleconferencing and ubiquitous broadband, no matter where in the world you were. Netflix, Zoom and Starlink all exploited the new “work and play from anywhere” paradigm, and tech stocks were hot once more.
Next came the launch of ChatGPT and the boom in large language models that could understand prompts and generate content, be it text, code, images or video — generative AI (Gen AI). For the first time it seemed AI was actually useful, and the chatbots were quickly recruited into the workplace by creators and office staff alike. So much so that when ChatGPT went down, it was a global calamity.
Led by Microsoft and Apple, the tech giants fuelled the boom, with AI chip maker Nvidia being the only hardware manufacturer to catch this new wave in a big way. Then came a spate of new “AI PCs”, laptops with built-in neural processors and optimised for Gen AI tasks. Tech was once more exciting; everyone needed new machines to keep up with their colleagues, and hardware sales surged.
But like all hype cycles, this wave has subsided. Smart chatbots have become boring, limited in their usefulness, and their tendency to invent “facts” or spout nonsense has tarnished their image, despite obvious productivity gains. Treat Gen AI with caution is the common sense approach — and watch those ballooning energy bills.
That hasn’t deterred the tech giants at all. Coming down the pipe is a new wave of Agent AI. Smart systems that can leverage the pretrained models to execute on a goal, employing “machine reasoning” instead of machine learning, to autonomously and recursively step through self-directed tasks and processes to get things done. And learning from their mistakes.
Best of all, Agent AI is designed for efficiency rather than brute force, with new systems-on-a-chip that optimise hardware use to get the most operations per watt, rather than per second, bringing smarts back to your laptop and phone, putting you in charge, and less dependent on energy-hungry datacentres and gigaband networks.
So, say hello to your new “Agent Smith”. It’s the future of your personal Matrix. / First published in Mindbullets June 6 2024.
- Despite appearances to the contrary, Futureworld cannot and does not predict the future. The Mindbullets scenarios are fictitious and designed purely to explore possible futures, and challenge and stimulate strategic thinking.
Would you like to comment on this article?
Sign up (it's quick and free) or sign in now.
Please read our Comment Policy before commenting.