From Determinism Toward Probabilism: A New Culture of Probability

We Really Had It Good!

There was a time when things were clear. An image served as proof of a particular event, a poetic text was evidently the product of a few artistic minds, and a Picasso stood out unmistakably among a thousand other cubist paintings. But today, that’s no longer the case. Things began to change as early as 2016.

“The Next Rembrandt” was a project developed by ING, Microsoft, Delft University of Technology, and other partners, aimed at creating a new painting in the style of Rembrandt using artificial intelligence and 3D printing. By analyzing the Dutch master's works with machine learning algorithms, the team generated an original painting that faithfully mimicked his style. The goal was to explore the boundary between art and technology. The result was astonishing (see image). At that time, we weren’t yet talking about Transformers (Google, 2017) — which now underpin the most advanced AI models — but instead, early diffusion systems that carved images out of noise (like the static on an old CRT television with no signal). Even art critics acknowledged that The Next Rembrandt could genuinely pass as a work by Rembrandt himself.

The focus today has shifted to a new question: How can we tell whether an event is real or fake?

Our human eyes can't reliably distinguish a real painting from a forged one. Our ears can't tell the difference between a naturally produced sound and one generated synthetically. Our minds are easily deceived by what we see and hear. So how can we protect ourselves?

We believe AI can also play a crucial role in defense. It won’t be our eyes, but artificial ones that we build to protect us—like sunglasses shielding us from intense light, IdentifAI - Find Origin - de-generative models will safeguard human perception from realities that are not entirely true. These "glasses" are powered by artificial intelligence—an antagonistic AI built to detect and confront generative AI.

Determinism

Determinism is a philosophical concept stating that every event, action, or decision is the inevitable result of preceding causes, following a logical and necessary chain. In this view, the universe operates like a perfect machine, where every cog follows a prewritten path. Thinkers such as Spinoza and Laplace believed that if we knew all the initial conditions of a system, it would be theoretically possible to predict every future event. In everyday life, determinism shapes our expectations of consistency and predictability: we expect the sun to rise, that every action has a consequence, and that rules work in a consistent way.

Probabilism

Probabilism, by contrast, accepts uncertainty as a fundamental feature of reality. It’s not merely about ignorance or lack of data—it’s the idea that some events can only occur with a certain probability. This paradigm is crucial today in fields like cybersecurity and finance, where threats and behaviors cannot be predicted with certainty, only estimated in terms of risk. Predictive algorithms, statistical analysis, and stochastic models reflect a probabilistic approach: acknowledging that we can’t know everything, but we can still act rationally based on probabilities.

Reality in the Age of AI

Artificial intelligence adopts a probabilistic rather than deterministic approach because it operates in complex, uncertain environments full of variables that are not always known or controllable. Unlike deterministic systems, which require strict rules and complete initial conditions to yield precise outcomes, AI works with data that is often incomplete, noisy, or ambiguous. Therefore, it relies on statistical and machine learning models that don’t deliver absolute truths but estimate the likelihood that a given hypothesis is correct. This allows AI to adapt to context, update predictions with new information, and make effective decisions even in uncertain scenarios. For instance, a facial recognition system doesn’t “decide” with certainty who a person is—it assigns probabilities to each possible match. In this sense, probabilism isn’t a limitation, but an evolved strategy for dealing with the complexity of reality.

AI Detection Technologies

Just like generative AI models, IdentifAI is built on probabilistic architectures—algorithms and neural networks that don’t follow deterministic logic but make statistical inferences. Within this framework, the very concept of “absolute accuracy” loses its meaning: we can’t state with irrefutable certainty whether a piece of content is natural or artificial—only assign a probability, which may shift over time as generative technologies evolve and the technological landscape changes. This is not a weakness, but the very essence of modern AI: a dynamic, adaptive, continuously learning technology.

But the reflection goes beyond any single tool. As AI becomes massively and pervasively integrated into our daily lives—as is already happening—we can’t limit ourselves to using it passively. We must rethink our decision-making paradigms—whether in business, science, or society. Deterministic models, rooted in linear cause-effect logic and claims of absolute truth, will begin to reveal their limits. In a world increasingly run by intelligent systems that produce forecasts, hypotheses, correlations, and error margins, human processes, too, will need to adopt a probabilistic mindset.

We will have to develop a culture of probability, of uncertainty and risk management—not as a surrender of responsibility, but as a mature and conscious acknowledgment that decisions must be made in a world that is inherently complex and never entirely predictable.

In this scenario, everything changes: How businesses plan, how scientists validate theories, how governments make policy decisions—and even how individuals form opinions. The shift from determinism to probability won’t be merely technical—it will be profoundly cultural. And preparing for this shift will be essential if we are to seize the opportunities of AI without being overwhelmed by them.

Text translated from Italian by AI.