Dr. Karl Friston is a distinguished computational psychiatrist, neuroscientist, and pioneer of modern neuroimaging and, now, AI. He is a leading expert on intelligence, natural as well as artificial. I have followed his work as he and his team uncover the principles underlying mind, brain, and behavior based on the laws of physics, probability, causality and neuroscience.

In the interview that follows, we dive into the current artificial intelligence landscape, discussing what existing models can and can’t do, and then peer into the divining glass to see how true artificial consciousness might look and how it may begin to emerge.

Current AI Landscape and Biological Computing

GHB: Broadly speaking, what are the current forms of AI and ML, and how do they fall short when it comes to matching natural intelligence? Do you have any thoughts about neuromorphic chips?

KF: This is a pressing question in current AI research: should we pursue artificial intelligence on high performance (von Neumann) computers or turn to the principles of natural intelligence? This question speaks to a fork in the road ahead. Currently, all the money is on artificial intelligence—licensed by the truly remarkable competence of generative AI and large language models. So why deviate from the well-trodden path?

There are several answers. One is that the artificial path is a dead end—in the sense that current implementations of AI violate the principles of natural intelligence and thereby preclude themselves from realizing their ultimate aspirations: artificial general intelligence, artificial super intelligence, strong AI, et cetera. The violations are manifest in the shortcomings of generative AI, usually summarized as a lack of (i) efficiency, (ii) explainability and (iii) trustworthiness. This triad neatly frames the alternative way forward, namely, natural intelligence.

So, what is natural intelligence? The answer to this question is simpler than one might think: natural intelligence rests upon the laws or principles that apply to the natural kinds that constitute our lived world. These principles are readily available from the statistical physics of self-organization, when the notion of self is defined carefully.

Put simply, the behavior of certain natural kinds—that can be read as agents. like you and me—can always be described as self-evidencing. Technically, this entails minimizing self-information (also known as surprise) or, equivalently, seeking evidence (also known as marginal likelihood) for an agent’s internal model of its world². This surprise is scored mathematically with something called variational free energy.

The model in question is variously referred to as a world or generative model. The notion of a generative model takes center stage in any application of the (free energy) principles necessary to reproduce, simulate or realize the behavior of natural agents. In my world, this application is called active inference.

Note that we have moved beyond pattern recognizers and prediction machines into the realm of agency. This is crucial because it means we are dealing with world models that can generate the consequences of behavior, choices or actions. In turn, this equips agents with the capacity to plan or reason. That is, to select the course of action that minimizes the surprise expected when pursuing that course of action. This entails (i) resolving uncertainty while (ii) avoiding surprising outcomes. The simple imperative— to minimize expected surprise or free energy—has clear implications for the way we might build artifacts with natural intelligence. Perhaps, these are best unpacked in terms of the above triad.

Efficiency. Choosing the path of least surprise is the path of least action or effort. This path is statistically and thermodynamically the most efficient path that could be taken. Therefore, by construction, natural intelligence is efficient. The famous example here is that our brains need only about 20 W—equivalent to a light bulb. In short, the objective function in active inference has efficiency built in —and manifests as uncertainty-resolving, information-seeking behavior that can be neatly described as curiosity with constraints. The constraints are supplied by what the agent would find surprising—i.e., costly, aversive, or uncharacteristic.

A failure to comply with the principle of maximum efficiency (a.k.a., principle of minimum redundancy) means your AI is using the wrong objective function. This can have severe implications for ML approaches that rely upon reinforcement learning (RL). In RL, the objective function is some arbitrary reward or value function. This leads to all sorts of specious problems; such as the value function selection problem, the explore-exploit dilemma, and more³. A failure to use the right value function will therefore result in inefficiency—in terms of sample sizes, memory requirements, and energy consumption (e.g., large language models trained with big data). Not only are the models oversized but they are unable to select those data that would resolve their uncertainty. So, why can’t large language models select their own training data?

This is because they have no notion of uncertainty and therefore don’t know how to reduce it. This speaks to a key aspect of generative models in active inference: They are probabilistic models, which means that they deal with probabilistic “beliefs”—about states of the world—that quantify uncertainty. This endows them not only with the capacity to be curious but also to report the confidence in their predictions and recommendations.

Explainability. if we start with a generative model—that includes preferred outcomes—we have, by construction, an explainable kind of generative AI. This is because the model generates observable consequences from unobservable causes, which means that the (unobservable or latent) cause of any prediction or recommendation is always at hand. Furthermore, predictions are equipped with confidence intervals that quantify uncertainty about inferred causes or states of the world.

The ability to encode uncertainty is crucial for natural intelligence and distinguishes things like variational autoencoders (VAE) from most ML schemes. Interestingly, the objective function used by VAEs is exactly the same as the variational free energy above. The problem with variational autoencoders is that they have no agency because they do not act upon the world— they just encode what they are given.

Trustworthiness: if predictions and recommendations can be explained and qualified with quantified uncertainty, then they become more trustworthy, or, at least, one can evaluate the epistemic trust they should be afforded. In short, natural intelligence should be able to declare its beliefs, predictions, and intentions and decorate those declarations with a measure of uncertainty or confidence.

There are many other ways we could unpack the distinction between artificial and natural intelligence. Several thought leaders—perhaps a nascent rebel alliance—have been trying to surface a natural or biomimetic approach to AI. Some appeal to brain science, based on the self-evident fact that your brain is an existence proof for natural intelligence. Others focus on implementation; for example, neuromorphic computing as the road to efficiency. An interesting technical issue here is that much of the inefficiency of current AI rests upon a commitment to von Neumann architectures, where most energy is expended in reading and writing from memory. In the future, one might expect to see variants of processing-in-memory (PIM) that elude this unnatural inefficiency (e.g., with memristors, photonics, or possibly quantum computing).

Future AI Development

GHB: What does truly agentic AI look like in the near-term horizon? Is this related to the concept of neuromorphic AI (and what is agentic AI)?

KF: Agentic AI is not necessarily neuromorphic AI. Agentic AI is the kind of intelligence evinced by agents with a model that can generate the consequences of action. The curiosity required to learn agentic world models is beautifully illustrated by our newborn children, who are preoccupied with performing little experiments on the world to see what they can change (e.g., their rattle or mobile) and what they cannot (e.g., their bedtime). The dénouement of their epistemic foraging is a skillful little body, the epitome of a natural autonomous vehicle. In principle, one can simulate or realize agency with or without a neuromorphic implementation; however, the inefficiency of conventional (von Neumann) computing may place upper bounds on the autonomy and agency of edge computing.

VERSES AI and Genius System

GHB: You are the chief scientist for VERSES AI, which has been posting groundbreaking advancements seemingly every week. What is Genius VERSES AI and what makes it different from other systems? For the layperson, what is the engine behind Genius?

KF: As a cognitive computing company VERSES is committed to the principles of natural intelligence, as showcased in our baby, Genius. The commitment is manifest at every level of implementation and design:

• Implementation eschews the unnatural backpropagation of errors that predominate in ML by using variational message-passing based on local free energy (gradients), as in the brain.
• Design eschews the inefficient top-down approach—implicit in the pruning of large models—and builds models from the ground up, much in the way that our children teach themselves to become autonomous adults. This ensures efficiency and explainability.
• To grow a model efficiently is to grow it under the right core priors. Core priors can be derived from first principles; for example, states of the world change lawfully, where certain quantities are conserved (e.g., object permanence, mathematical invariances or symmetry, et cetera), usually in a scale-free fashion (e.g., leading to deep or hierarchical architectures with separation of temporal scales).
• Authentic agency is assured by equipping generative models with a minimal self-model; namely, “what would happen if I did that?” This endows them with the capacity to plan and reason, much like System 2 thinking (planful thinking), as opposed to the System 1 kind of reasoning (intuitive, quick thinking).

At the end of the day, all this rests upon using the right objective function; namely, the variational free energy that underwrites self-evidencing. That is, building the most efficient model of the world in which the agent finds herself. With the right objective function, one can then reproduce brain-like dynamics as flows on variational free energy gradients, as opposed to costly and inefficient sampling procedures that are currently the industry standard.

Consciousness and Future Directions

GHB: What might we look forward to for artificial consciousness, and can you comment on the work with Mark Solms?

KF: Commenting on Mark’s work would take another blog (or two). What I can say here is that we have not touched upon two key aspects of natural intelligence that could, in principle, be realized if we take the high (active inference) road. These issues relate to interactive inference or intelligence—that is, inference among agents that are curious about each other. In this setting, one has to think about what it means for a generative model to entertain the distinction between self and other and the requisite mechanisms for this kind of disambiguation and attribution of agency. Mark would say that these mechanisms rest upon the encoding of uncertainty—or its complement, precision —and how this encoding engenders the feelings (i.e., felt-uncertainty) that underwrite selfhood.

In the modern age, artificial intelligence (AI) is revolutionizing how we live, work, and think – sometimes in ways we don’t fully understand or anticipate. In newsrooms, classrooms, boardrooms, and even bedrooms, tools like ChatGPT and other large language models (LLMs) are rapidly becoming standard companions for generating text, conducting research, summarizing content, and assisting in communication. But as we embrace these tools for convenience and productivity, there is growing concern among educators, journalists, editors, and cognitive scientists that we are trading long-term intellectual development for short-term efficiency.

As a news editor, one of the most distressing observations has been the normalization of copying and pasting AI-generated content by young journalists and writers. Attempts to explain the dangers of this trend – especially how it undermines the craft of writing, critical thinking, and authentic reporting – often fall on deaf ears. The allure of AI is simply too strong: its speed, its polish, and its apparent coherence often overshadow the deeper value of struggling through a thought or refining an idea through personal reflection and effort.

This concern is not isolated to journalism. A growing body of research across educational and corporate environments points to an overreliance on writing tools as a silent threat to cognitive growth and intellectual independence. The fear is not that AI tools are inherently bad, but that their habitual use in place of human thinking – rather than in support of it – is setting the stage for diminished creativity, shallow learning, and a weakening of our core mental faculties.

One recent study by researchers at the Massachusetts Institute of Technology (MIT) captures this danger with sobering clarity. In an experiment involving 54 students, three groups were asked to write essays within a 20-minute timeframe: one used ChatGPT, another used a search engine, and the last relied on no tools at all. The researchers monitored brain activity throughout the process and later had teachers assess the resulting essays.

The findings were stark. The group using ChatGPT not only scored lower in terms of originality, depth, and insight, but also displayed significantly less interconnectivity between brain regions involved in complex thinking. Worse still, over 80% of students in the AI-assisted group couldn’t recall details from their own essays when asked afterward. The machine had done the writing, but the humans had not done the thinking. The results reinforced what many teachers and editors already suspect: that AI-generated text, while grammatically sound, often lacks soul, depth, and true understanding.

These “soulless” outputs are not just a matter of style – they are indicative of a broader problem. Critical thinking, information synthesis, and knowledge retention are skills that require effort, engagement, and practice. Outsourcing these tasks to a machine means they are no longer being exercised. Over time, this leads to a form of intellectual atrophy. Like muscles that weaken when unused, the mind becomes less agile, less curious, and less capable of generating original insights.

The implications for journalism are especially dire. A journalist’s role is not simply to reproduce what already exists but to analyze, contextualize, and interpret information in meaningful ways. Journalism relies on curiosity, skepticism, empathy, and narrative skill – qualities that no machine can replicate. When young reporters default to AI tools for their stories, they lose the chance to develop these essential capacities. They become content recyclers rather than truth seekers.

Educators and researchers are sounding the alarm. Nataliya Kosmyna, lead author of the MIT study, emphasized the urgency of developing best practices for integrating AI into learning environments. She noted that while AI can be a powerful aid when used carefully, its misuse has already led to a deluge of complaints from over 3,000 educators – a sign of the disillusionment many teachers feel watching their students abandon independent thinking for machine assistance.

Moreover, these concerns go beyond the classroom or newsroom. The gradual shift from active information-seeking to passive consumption of AI-generated content threatens the very way we interact with knowledge. AI tools deliver answers with the right keywords, but they often bypass the deep analytical processes that come with questioning, exploring, and challenging assumptions. This “fast food” approach to learning may fill informational gaps, but it starves intellectual growth.

There is also a darker undercurrent to this shift. As AI systems increasingly generate content based on existing data – which itself may be riddled with bias, inaccuracies, or propaganda – the distinction between fact and fabrication becomes harder to discern. If AI tools begin to echo errors or misrepresentations without context or correction, the result could be an erosion of trust in information itself. In such a future, fact-checking will be not just important but near-impossible as original sources become buried under layers of machine-generated mimicry.

Ultimately, the overuse of AI writing tools threatens something deeper than skill: it undermines the human drive to learn, to question, and to grow. Our intellectual autonomy – our ability to think for ourselves – is at stake. If we are not careful, we may soon find ourselves in a world where information is abundant, but understanding is scarce.

To be clear, AI is not the enemy. When used responsibly, it can help streamline tasks, illuminate complex ideas, and even inspire new ways of thinking. But it must be positioned as a partner, not a replacement. Writers, students, and journalists must be encouraged – and in some cases required – to engage deeply with their work before turning to AI for support. Writing must remain a process of discovery, not merely of delivery.

As a society, we must treat this issue with the seriousness it deserves. Schools, universities, media organizations, and governments must craft clear guidelines and pedagogies for AI usage that promote learning, not laziness. There must be incentives for original thinking and penalties for mindless replication. We need a cultural shift that re-centers the value of human insight in an age increasingly dominated by digital automation.

If we fail to take these steps, we risk more than poor essays or formulaic articles. We risk raising a generation that cannot think critically, write meaningfully, or distinguish truth from fiction. And that, in any age, is a far greater danger than any machine.

Please follow Blitz on Google News Channel

In a livestream on X, Musk both bragged about the model yet simultaneously fretted about the impact on humanity if the AI turns evil. 

“This is the smartest AI in the world,” said Musk while surrounded by members of his xAI team. “In some ways, it’s terrifying.”

He compared Grok 4 to a “super-genius child” in which the “right values” of truthfulness and a sense of honor must be instilled so society can benefit from its advances. 

Musk admitted to being “worried,” saying that “it’s somewhat unnerving to have intelligence created that is far greater than our own, and will this be bad or good for humanity?”

The xAI owner concluded that “most likely, it’ll be good.”

Musk said Grok 4 is designed to perform at the “post-graduate level” in many topics simultaneously, which no person can do. It can also handle images, generate realistic visuals and tackle complex analytical tasks.

Musk claims that Grok 4 would score perfectly on SAT and graduate-level exams like GRE even without seeing the questions beforehand.

Alongside the model release, xAI introduced SuperGrok Heavy, a subscription tier priced at $300 per month. A standard Grok 4 tier is available for $30 monthly, and the basic tier is free.

OpenAI, Google, Anthropic and Perplexity have unveiled higher-priced tiers as well: ChatGPT Pro, at $200 a month; Gemini Ultra, at $249.99 a month; Claude Max, at $200 a month; and Perplexity Max, for $200 a month.

See also: Elon Musk Startup xAI Launches App Offering Access to Grok Chatbot

Turbulent Week for Grok and X

Grok 4’s launch follows a turbulent week marked by antisemitic content generated by Grok 3 and the resignation of Linda Yaccarino, the CEO of X.

Grok 4 is being released in two configurations: the standard Grok 4 and the premium “Heavy” version.

The Heavy model features a multi-agent architecture capable of collaborative reasoning on challenging problems.

The model demonstrates advances in multimodal processing, faster reasoning and an upgraded user interface. According to xAI, Grok 4 can solve complex math problems, interpret images — including scientific visuals such as black hole collisions — and perform predictive analytics, such as estimating a team’s odds of winning a championship. 

Benchmark data shared by xAI shows that Grok 4 Heavy outperformed previous models on tests such as Humanity’s Last Exam.

xAI outlined an aggressive roadmap for the remainder of 2025: launching a coding‑specific AI in August, a multimodal agent in September and a model capable of generating full video by October.

Grok 4’s release intensifies the competition among leading AI firms. OpenAI is expected to roll out GPT‑5 later this summer, while Google continues to develop its Gemini series.

Read more: