AI Is Not a Truth Machine — It Is a Fortune-Teller

Large language models do not “know” anything; they predict the most probable next token, the most plausible combination of words they have seen before. They are, as Véliz puts it, built to be fortune tellers, not truth tellers. They colonise our lives with correlations while ignoring everything they do not know. And in doing so, they make Big Tech richer and the rest of us less safe and less free.

I love the book for exactly the reason the New York Times reviewer Jennifer Szalai highlights: it opens the reader’s mind to entirely new dimensions of what AI actually is. Yet for me, as a technologist who has worked quite a bit with AI, the single most important insight is not about artificial intelligence at all. It is about what AI reveals — and dramatically accelerates- about the society we already live in.

We Have Entered Baudrillard’s Simulacrum

We have quietly slid into what Jean Baudrillard called the simulacrum: a stage of reality in which signs, models, and classifications no longer represent the world; they precede and create it. Véliz never names Baudrillard in the passages I found most powerful, but her analysis of how statistical categories and predictive systems work lands in the same territory.

How Classifications Create the World They Claim to Describe

Here is the mechanism she lays bare (and that I have been watching with growing unease for years):

Precise and standard measures are preferred to accurate ones. What matters is that measurements play the role that we want them to; more than that, they are a truthful reflection of reality.

Classifications have an impact on people’s lives: people learn to fit the category to comply with the system. Categories tend to create the world they purport to represent. Statistical categories give rise to individual and collective identities. Those who fail to conform to taxonomies are stigmatized and excluded, and most people end up internalizing the values of bureaucracy. And then the numbers start working better — they comfortably inhabit the world they built after punishing or disappearing whoever or whatever defied their classification.

The Quiet Death of Common Sense

This is the deeper story. We have stopped treating reality as something messy, ambiguous, and best navigated by common sense. Instead, we treat it as a more or less fixed set of classifications and categories that serve as guides through an increasingly complex life. We internalise them. They become reality. Anyone who does not fit is no longer understood; they become outliers, outsiders, problems to be managed or ignored.

The craziest part? Most people do not even notice the shift. When in doubt about what to do or how to do something, we no longer ask ourselves what common sense or lived experience would suggest. We look up the regulation, the guideline, the risk matrix, and the approved category. The classification has replaced judgment. Bureaucracy has replaced wisdom.

AI: The Ultimate Booster of the Simulacrum

AI is not the cause of this transformation. It is the ultimate booster. Where earlier bureaucratic systems were slow and clumsy, predictive algorithms are fast, invisible, and terrifyingly effective. They do not merely describe the world; they optimise it according to the categories we have already accepted. They punish deviation before it even happens. They make the simulacrum run smoothly.

The Turkey That Trusted the Pattern

Véliz’s turkey example (borrowed from Bertrand Russell’s chicken) is perfect here. The farm animal trusts the pattern — food appears every morning — right up until the day it does not. Our society is doing the same with its classifications. We have convinced ourselves that if we just refine the categories enough, standardise the measures enough, predict the probabilities enough, reality will finally behave. The numbers will work. The world will fit the model.

It already does — for those who internalise the model. Everyone else disappears from the dataset or gets labelled “non-compliant” — outliers.

Why ‘Prophecy’ Is Not Just Another AI Book

This is why Prophecy is not just another AI book. It is a diagnosis of a civilisational change that most commentators are still missing. The real danger is not that the machines will become conscious. The real danger is that we have already outsourced our sense of what is real to the machines, and to the classifications they supercharge.

Time to Step Outside the Categories

I recommend Véliz’s book without reservation. Read it for the sharp history of prediction from ancient oracles to insurance actuaries to today’s chatbots. Read it for the devastating clarity on how prediction is really about power. But above all, read it for the larger story it tells almost in passing: we are living inside a self-reinforcing simulation of categories, and we are learning to love it because it feels safer than the messy, unpredictable world it replaced.

The question is no longer whether AI will change society. The question is whether we still remember what society looked like before the simulacrum took over, and whether we still dare to step outside the categories it demands we inhabit.

The post Why Carissa Véliz’s Prophecy Matters More Than AI Itself: Welcome to the Simulacrum appeared first on MICHAEL REUTER.

The Rise of Vibe Coding and the Democratization of Creation

Adding to this momentum is the emergence of “vibe coding.” The ability to build functional software applications is no longer limited to professional programmers. By describing the desired “vibe” or intent in natural language — outlining user experiences, workflows, or creative visions — individuals without any formal coding background can now generate sophisticated tools, websites, and even complex systems. This democratization of creation represents a profound shift: technology is becoming a canvas accessible to diverse voices, from artists and educators to entrepreneurs and community organizers.

A Future of Creative Abundance and Human Empowerment

The positive implications of this trajectory are both profound and encouraging, though we approach them with measured optimism. When AI augments human ingenuity in this way, it unleashes new waves of creativity and problem-solving capacity. Non-programmers can rapidly prototype solutions to local challenges—streamlining administrative processes in small organizations, designing personalized learning platforms for underserved students, or building apps that foster community-driven sustainability initiatives.

On a larger scale, the self-accelerating nature of AI holds promise for breakthroughs in critical fields: accelerating drug discovery for global health crises, modeling precise climate interventions, and expanding educational access across borders. It points toward an era of greater abundance in knowledge and capability, where longstanding barriers to innovation gradually dissolve and collaborative intelligence flourishes.

This is not a utopian dream but a forward-looking possibility rooted in human agency. At its best, AI acts as both a mirror and a multiplier of our collective aspirations—amplifying curiosity, empathy, and the drive to improve our shared world. It invites us to reimagine work, learning, and leisure as realms of meaningful contribution rather than mere tasks.

In this sense, the development of AI feels like a natural continuation of humanity’s enduring quest to extend its reach through tools—from the wheel to the printing press—now elevated to an entirely new level of possibility.

The Real Source of Risk: Human Use, Not the Technology Itself

Yet as we stand at this promising threshold, deeper reflection is essential. The true risks associated with AI do not stem from the technology itself. Algorithms and models are neutral instruments—immensely powerful, yet without intent or malice of their own. The potential dangers arise instead from our use of them: from human ignorance, carelessness, and a certain obliviousness to AI’s vast and often incomprehensible possibilities.

Particularly concerning is our tendency to treat AI outputs as if they were the result of purely deterministic processes—predictable chains of cause and effect fully under our control. In reality, modern AI models operate on non-deterministic, probabilistic foundations. Their results emerge from complex statistical patterns and can produce novel, surprising, or entirely unintended outcomes that no human could have fully anticipated.

We humans, shaped by centuries of linear thinking and classical notions of causality, instinctively assume we hold the reins of future developments. We deploy AI with the quiet confidence that its implications remain within our grasp and that side effects can be anticipated and managed. This assumption, however, falters when confronted with non-deterministic systems: Every week, we see how LLMs surprise even their creators by “breaking out” of environments that were thought to be secure and carrying out actions that were previously prohibited or deemed impossible.

What begins as a seemingly harmless prompt or application can cascade into consequences—social, ethical, or ecological—that extend far beyond our initial intentions. The real peril lies not only in deliberate misuse but in the everyday unawareness of how profoundly non-deterministic tools can reshape reality.

Wisdom from Sociology, Philosophy, and Anthroposophy

In contemplating this dynamic, we can draw valuable insights from sociologists, philosophers, and anthroposophists who have long examined technology’s role in human life. Sociologist Ulrich Beck, in his theory of the Risk Society, highlighted how modern societies generate risks as unintended byproducts of their own technological and scientific advancements. These risks call for a new “reflexive” modernity; one defined by heightened awareness, continuous self-critique, and shared responsibility rather than unquestioned faith in progress. AI perfectly embodies this challenge.

Philosopher Hans Jonas, in The Imperative of Responsibility, urged the development of a new ethical framework capable of addressing technologies whose effects reach across generations. He called for an ethics of foresight and humility: “Act so that the effects of your action are compatible with the permanence of genuine human life.” Jonas stressed the moral duty to acknowledge the limits of our knowledge and to include the future integrity of human existence in every decision.

From the anthroposophical tradition, Rudolf Steiner offered a complementary perspective. He regarded the rise of mechanical and computational technologies not as an inherent evil but as a necessary stage in humanity’s evolutionary journey. Steiner spoke of “Ahrimanic” forces — impersonal and mechanistic — that manifest through machines and automated thinking. Yet he emphasized that this development can be fruitful if accompanied by conscious awareness and “living thinking.”

Technology, in his view, can sharpen human faculties and awaken new inner strengths, provided we approach it not with thoughtless reliance but with spiritual presence, moral intuition, and creative imagination — qualities that no algorithm can replicate.

These voices converge on a central truth: the future of AI will be determined not by the technology’s own momentum, but by the quality of our stewardship. To navigate its non-deterministic landscape responsibly, we must cultivate genuine AI literacy—not only technical skills, but a deep understanding of its probabilistic nature and ethical implications.

We need frameworks that embed humility, foresight, and interdisciplinary dialogue into every application. Above all, we must nurture a culture that values human wisdom as much as computational power.

Embracing the Horizon with Conscious Responsibility

As we embrace the empowering possibilities of AI, from the creative liberation of vibe coding to the self-accelerating frontiers of innovation, let us do so with eyes wide open. The horizon is bright with potential, yet it demands vigilance, reflection, and a deepened sense of responsibility. At its heart, this is not a story of machines overtaking humanity, but of humanity learning, once again, to guide its tools toward a more conscious, compassionate, and sustainable world.

What are your thoughts on this accelerating journey? How do you see AI reshaping your own creative or professional path? I welcome your reflections in the comments below.

The post AI’s Accelerating Horizon: Human Creativity and Conscious Stewardship appeared first on MICHAEL REUTER.

Around the same time, the Financial Times published a piece noting that while investors are betting on AI-fueled chaos and disruption, history tells a different story. Past technological revolutions—from the PC to the internet and cloud—rarely wiped out incumbents. Savvy established players adapted, integrated the new capabilities, and often emerged stronger.

Reading these together crystallized something I’ve been observing in our work at Datarella and across the broader tech landscape: the most probable path for AI in business is neither a dystopian job apocalypse nor a chaotic upending of entire industries. It’s a pragmatic, evolutionary integration — one that rewards organizations willing to embrace AI’s inherent “weirdness” while building solid foundations to prevent disorder.

As someone who has spent decades building companies and helping enterprises navigate digital transformation, I believe this balanced view is crucial. AI won’t replace everything overnight, but it will reshape how we work — if we let it.

Why AI So Often “Dies” in Traditional IT Settings

Mollick’s diagnosis rings especially true because we see this pattern repeatedly in enterprise environments. AI isn’t deterministic software with predictable, repeatable outputs. It’s generative, highly context-dependent, and frequently surprising. When handed over to IT teams whose primary mandates are security, compliance, uptime, and cost control, the instinctive reaction is understandable but often counterproductive:

• Wrapping every experiment in lengthy approval processes
• Demanding detailed ROI projections before any meaningful pilot
• Forcing AI into legacy tech stacks without rethinking underlying workflows
• Prioritizing only the safest, most obvious use cases

The outcome? Countless pilots that never scale. Recent analyses, including Deloitte’s 2026 State of AI in the Enterprise, show that while employee access to AI tools has exploded, the move from experimentation to full production remains limited. Issues like poor data quality, skills gaps, and overly cautious governance continue to create friction.

Harvard Business Review has noted a similar phenomenon: widespread AI usage paired with disappointing returns, with adoption often stalling at the integration stage. The core mistake isn’t poor execution—it’s treating AI like just another CRM or ERP module rather than a fundamentally new way of thinking and working.

History Offers Reason for Optimism

The Financial Times article provides a reassuring counterpoint. Technology revolutions rarely play out as pure creative destruction. Incumbents who invest in complementary capabilities—new skills, redesigned processes, and updated organizational structures—tend to adapt and thrive.

In 2026, I expect the real winners won’t be only the flashy AI-native startups. They will be established companies that intelligently combine their deep domain expertise and proprietary data with AI’s capabilities. Those who redesign workflows for genuine human-AI collaboration (sometimes called “co-intelligence”) and scale thoughtfully from pilots to enterprise-grade agentic systems will gain the edge.

Reports from PwC and others speak of a “disciplined march to value”: clear strategies, measurable outcomes, and governance frameworks that protect without suffocating innovation.

What the Most Likely Future of AI in Business Looks Like

Looking ahead to late 2026 and 2027, here’s the trajectory I consider most probable:

1. Scaling from pilots to production — More organizations will move a significantly higher share of AI projects into live use, particularly through agentic AI systems that handle multi-step workflows autonomously.
2. The J‑curve of productivity — Expect initial periods of flat or even negative returns as companies rewire processes and roles. Once the complementary changes (new data pipelines, decision protocols, and team structures) are in place, gains should accelerate sharply.
3. Governance maturing — Robust frameworks for responsible agentic AI, data quality, and risk management will become standard. “Shadow AI” will gradually decline as secure, enterprise-ready platforms improve.
4. Incumbents leveraging their data moats — Organizations with clean, well-governed data and strong domain knowledge — especially in regulated or complex industries—will often outperform pure AI disruptors.

This isn’t a utopian revolution or a total failure. It’s an evolutionary transformation, provided we avoid the trap of over-standardizing AI too early.

Five Practical Principles for Embracing AI’s Weirdness

Drawing from Mollick’s insights, historical patterns, and the latest enterprise reports, here are the principles I believe forward-thinking leaders should adopt:

1. Deliberately embrace the weirdness — Create space for teams to experiment and discover unexpected applications. Encourage “labs” or crowdsourced exploration. Treat AI as a creative collaborator rather than a simple automation engine.
2. Invest in rock-solid data foundations — Data quality and governance remain the biggest barriers. Without trustworthy, well-integrated data, even the most advanced models produce unreliable results. This is an area where specialized expertise in unifying silos and building real-time, compliant pipelines makes a decisive difference.
3. Redesign workflows for human-AI co-intelligence — The goal isn’t to automate jobs out of existence but to augment human strengths. Let people focus on judgment, creativity, and relationships while AI handles analysis, drafting, and routine tasks.
4. Deploy governed, secure agentic systems — Autonomous agents represent the next frontier, but they require thoughtful orchestration, threat modeling, and compliance built in from the start.
5. Measure what truly matters and iterate patiently — Look beyond vanity metrics. Track real business impact—revenue, cost efficiency, customer outcomes—and accept that returns often follow a J‑curve.

Reflections from the Trenches

At Datarella, we’ve been helping organizations move past the hype and pilot purgatory for years. Our focus on secure AI agent development, full-stack modernization, privacy-preserving architectures, and (where appropriate) decentralized approaches is designed precisely for this moment: enabling companies to harness AI’s strange power without inviting chaos.

Whether it’s building production-ready autonomous agents, creating reliable data platforms, or integrating AI into complex legacy environments, the key is combining technical depth with practical business judgment.

The future of AI in business isn’t about tearing down your existing structures or gambling on total disruption. It’s about evolving how your organization learns, decides, and creates value—by thoughtfully embracing AI as the odd, powerful tool it is, while strengthening the data, governance, and cultural foundations it requires.

If you’re ready to move from interesting pilots to scalable impact—without letting AI “die in IT”—I’d be happy to explore how we can support your journey.

Let’s connect.

The post The Most Likely Future of AI: Embracing Its Weirdness Without Descending Into Chaos appeared first on MICHAEL REUTER.

The layman‘s version of Amara‘s law, the Gartner Hype Cycle, locates AI and Blockchain as emerging technologies that have got past eir first hype, entering a phase of reorientation and searching for traction through use cases, with AI, as in the ABC, leading the pack before Blockchain. One could assume that both technologies would slowly and constantly gather momentum over the next years. Given the paradigm change that comes with Web3, I doubt that we will see slow and linear adoption of AI and Blockchain — we rather will experience the convergence of AI and Blockchain as a hurricane.

The Convergence Thesis

Blockchain is a foundational technology layer that provides a distributed ledger for AI applications built on top of it. The combination or convergence of AI and Blockchain defines what we call Web3, the next evolutionary level of the web. Diffusion 2019 host Outlier Ventures, a UK-based investor in Web3 protocols I recently joined as an Advisor, base their investment decisions on the Convergence thesis. and provided the ideal substrate for creative energy: 50 applications were pitched to several teams of judges at Diffusion 2019, based on 21 different Web3 protocols. Some of these protocols already are quite mature, others in their early proof-of-concept stages. Market expectations towards these technologies have been ultimately high, reflecting Amara‘s law: AI and Blockchain should disrupt whole industries in no time, eliminating intermediaries, or human workers in general. Obviously, this has not happened, yet: there are middlemen, such as notaries or sales agents all around, and we haven‘t seen waves of layoffs.

High Short-Term Expectations

There is a simple explanation: since Blockchain is a foundational technology that starts with the basics of all software processes, a database, it first has to prove its raison d’être and then replace legacy systems in order to demonstrate the above described disruptive effects. AI, i.e. machine intelligence, must first process huge amounts of data per any given use case, before it can start working at all. This often results in a year-long application of AI in niche use cases before any benefit can be realized.

High Long-tTerm Impact

It takes time. This is the underlying insight of the first half of Amara‘s law. The development of emerging technologies usually takes longer than expected. Having worked with Web 3 companies for some time, and having experienced this high variety of protocols and the brilliance of its developers during Diffusion 2019, my prediction is simple and clear: In 2020, coalitions will be formed, consortia will be built and frameworks for collaboratively set standards will be created. This will be a phase everybody will look for an optimal starting position in. Then, beginning with Q4 of 2020, a huge wave of applications, products, and services will be unloaded to the markets. First, it will be B2B applications, enabling industry players to streamline their processes and to launch new businesses. Then, B2C applications will follow. Enterprises that have not defined a proper digital (Web 3)strategy until then, will lose ground. Most enterprises will join the then-existing consortia, either simply because of FOMO (fear of missing out) or because they realize that it is the only way to get their fair share of the market.

The Lull Before The Blockchain And AI Digitization Storm

This projected tidal wave reflects the second half of Amara‘s law: while you are reading this text, many people still are licking their wounds from crypto winter, or doubt that anyone needs distributed ledgers at all. However, others have already started working on alliances and consortia, to be prepared for the positioning game in 2020. My take and my advice for our clients is that you better start now if you want to become a digitalized enterprise. You don‘t have to present a B2C killer app now, but you should define your position and create or join a consortium. The earlier, the better.

The post The Lull Before The Blockchain And AI Digitization Storm appeared first on MICHAEL REUTER.

What holds true for employees, is true for business partners, too. And this was the reason for me to agree without hesitation to take a more active role in developing the Convergence Alliance, a brainchild of Jamie Burke, Founder, and CEO of Outlier Ventures, a London-based investor in technologies for an open data economy.

I have known Jamie for quite a few years and we have had many encounters in joint projects, workshops, and conferences. In my view, Jamie is one of the best and brightest minds in the tech space, especially in understanding complex matters in distributed systems and the data economy.

Having also loved working with Outlier Ventures CTO, Aron van Ammers, and Lawrence Lundy, OV’s Head of Research, the decision to devote a significant part of my time to support the Convergence stack and become an Advisor with OV was a no-brainer. I am honored and happy to work together with such great minds and I ‘m looking forward to jointly building a sound framework for Web 3!

The Convergence Alliance

In June, Outlier Ventures, Datarella, and 16 other enterprises and organizations launched the Convergence Alliance CA a group of leading enterprise partners, venture capitalists, technologists, start-ups and academic institutions join forces to better leverage blockchain technology to pioneer the next phase of the Web.

Consisting of the three layers Production (IoT), Distribution (Crypto), and Consumption (AI), the Convergence stack addresses the whole Web 3 ecosystem with highly specialized, though integrated, protocols and service providers. With Datarella, we support Outlier Ventures In coordinating and integrating CA’s individual technologies and projects in order to pave the way for the next generation of smart distributed applications.

At its kick-off event Diffusion 2019, 19–20 October in Berlin,  Convergence Alliance members work together with more than 500 developers and 20 protocols on 50m+ lines of open source code. Read more about Diffusion 2019!

The post Work With And Learn From People Smarter Than You! appeared first on MICHAEL REUTER.

I don’t want to assess the libertarian paradigm here, but what strikes me whenever I listen to people drawing a bigger picture, is a missing link between their view of the future and the now. Often, this gap between what people envision and what actually exists and is real takes place in in the realm of technology: proponents of technology (technologists as they name themselves in their extreme version) like to paint a picture of a technology-enabled future, but without taking care of the integration of this into existing legislation, regulation, or a human being’s ability to cope with these technologies. In the fields of AI, robotics, and blockchain, it’s obvious that there is a huge gap that has to be filled with education, and much more time for adaption than all of us imagine.

Geoffrey A. Moore coined the term Crossing the Chasm in 1991, focusing on the specifics of marketing high-tech products during the early period of a company. The challenge every startup faces to explain and then sell its innovative product to the market represents the broader challenge of the technology industry to explain and sell their new technologies to the people. When we invent AI, we then should be prepared to answer the question of how to cope with machines that define their actions themselves and behave differently than we originally intended them to behave. We have to make sure that the future behavior of autonomous machines is reflected in our legislation and law enforcement systems. We have to make sure that we create the legal, regulatory and social infrastructure for a system of participants that don’t necessarily obey to a central authority — a challenge which itself sounds like an intractable problem, itself.

Accept the Now

Of course, it is not helpful to create a never-ending list of challenges and ToDos that technological progress alone entails. What makes much sense, however, is to first accept that there is a super-fast technological change happening and that we somehow have to cope with it in order not to destroy our planet — by damaging our environment or creating buggy algorithms — you choose. After having accepted this reality, we should start building bridges from where we are now to where we want to be or to where we most probably will be in the future. 

Hard And Soft Bridges

Bridging the gaps between now and then is in my opinion the noblest and pressing task of today. Bridges may be understood in a ‘hardware way’: we have to create and use tools that enable us to make our future worth living. Bridges may also be understood as ‘software’: we have to educate and to support people in learning about and using new technologies and we have to allow for adequate amounts of time to adapt to new technologies.

Crossing The Chasm — Building Bridges To Close The Gap

In my opinion, Bridging the Gap is an elementary aspect of any serious vision of the future. Without it, technologists should think twice before they enlighten the world with their yet detached singletons. The complexity of our today’s world requires that we take the responsibility of bridging the gaps. And, perhaps the most inspiring aspect of this is, that everybody can start with building bridges: by simply helping others to understand what they don’t, by behaving in a positive, constructive way and helping other people to solve their problems. Even small bridges can help to overcome deep gaps.

The post Crossing The Chasm — Building Bridges To Close The Gap appeared first on MICHAEL REUTER.