Nilesh Jasani: Get set for a world of ever-evolving super-Einsteins

Then came the current Agentic Era. No longer satisfied with talk, we sought action. Artificial intelligence (AI) has begun booking flights, editing selfies, navigating spreadsheets and doing several daily tasks. These early agents, while powerful, remain constrained—they are brilliant assistants, but still locked in their digital cribs.

Also Read: Dave Lee: Apple must make peace with developers for AI success

Yet, something far more transformative lies ahead. The third stage in this journey will see intelligence untethered from digital devices. This is when cognition escapes the screen and begins to permeate the physical world. Whether called embodied AI, robotics or the ‘era of smart everything,’ this phase will bring adaptive learning systems into everything from fork-lifts to furniture. Powered by action models, experience learning, multi-modal understanding and advanced hardware, machines will begin to learn from and reshape the world around them—physically, not just virtually.

And even this would only be a warm-up. The fourth stage promises an intelligence explosion. We are rapidly approaching an era where the most complex and longstanding human challenges will be met with cognitive power vastly exceeding our own. Some AI models are already rivalling Olympiad-level students in mathematics. It is a matter of time before these systems surpass the most brilliant human minds in every discipline.

This intelligence, endlessly scalable and tirelessly improving, will first prove its worth in the realm of health.

Also Read: AI, identity and drama: Why everyone’s turning into a character

Smarter diagnostics as an early sign of Stage 4 success: While public fascination remains fixated on humanoid robots, self-driving cars and laundry-folding machines, the real transformation is already underway—in diagnostics. Here, AI has begun to outperform human experts in identifying disease from X-rays, cancer scans and medical imagery. These are not just marginal improvements. They are leaps in precision, speed and scalability.

This diagnostic revolution is more than a healthcare upgrade. It signals AI’s capacity to reason through complexity in ways that surpass even expert human cognition. If an algorithm can outperform trained radiologists, it suggests a broader capability to interpret, hypothesize and solve problems. These are cognitive feats previously limited to specialists. Now, machines are taking them on—and winning.

Even if some of these breakthroughs still await peer-reviewed confirmation, the trend is unmistakable. Diagnostic models are showing that AI can attack complexity head-on, making decisions that would take humans days or weeks, within seconds. And diagnostics is only the beginning.

Also Read: When AI gets a manager, you know the game has changed

Drug discovery will be a bigger test: If diagnostics is about pattern recognition, drug discovery demands generative intelligence. It involves not only spotting problems, but imagining novel solutions, designing new molecules and validating hypotheses through layers of experimentation.

Historically, biology has been structured around frameworks that made sense to humans—grouped proteins, labelled pathways, hierarchical taxonomies, etc. These simplifications helped us manage biochemical complexity but fell short of describing reality in full. Machines, however, are not bound by cognitive short-cuts. They operate across molecular spaces and mechanistic landscapes too vast for humans to hold in their mind.

AI in drug discovery now proposes ideas, runs simulations and offers predictive insights across thousands of dimensions. Tools like Absci’s zero-shot antibody generators can design viable drug candidates without needing examples. Recursion’s phenomics platform screens tens of thousands of compound-cell combinations simultaneously. These capabilities hint at a new paradigm: one where machines don’t just assist in discovery, they drive it.

The real story isn’t faster timelines or cheaper trials. It’s the radical expansion of what’s possible. AI is allowing science to ask more questions, explore more hypotheses and navigate a vastly larger solution space than ever before.

Also Read: The agentic AI revolution isn’t the future, it’s already here

This is only the beginning: What we see today is just the surface. Beneath it lies a revolution in how we interact with chemistry, biology and the material world. Synthetic biology is already leveraging AI to design gene circuits that function correctly on their first attempt. Antibody design is being reshaped by systems that require no training data. Whole-cell simulations are on the horizon. Predictive models now anticipate binding affinity, structural stability and biological impact without the need for physical experimentation.

These shifts are not incremental; they’re multiplicative. Each advance unlocks others. Taken together, they transform science from a linear process into something exponential.

Beyond the spectacle: Today’s headlines remain preoccupied with theatrical AI feats—machines that draft emails, pass exams, compose jingles or book your dinner reservation without being prompted. Entertaining, yes. Useful, perhaps. But these tasks belong to an ageing class of applications.

The real story is elsewhere: We are not simply entering a world of better tools. We are entering a world with new minds—artificial ones that can reason in ways foreign to our own. These systems will not just support discovery; they will co-create it. Their thinking will be alien, powerful and deeply unfamiliar. And yet, increasingly, these tools will be indispensable.

Also Read: The great AI reboot: Educators, techies and leaders all need to adapt fast

At the same time, we are inevitably headed for complex terrain. Legal, ethical, moral, social and institutional questions—each deserving volumes of discussion—loom large. From accountability in autonomous systems to the governance of machine-generated knowledge, the implications are vast and underexplored. This article cannot do justice to those issues, but it can flag a simple truth: the genie is not going back in the bottle.

What matters now is whether various communities—scientific, industrial, governmental and educational—start preparing for what lies ahead. Because Generative AI is not, and will not be, mostly about chatbots and digital assistants. Those tools dominate today’s conversations, but they are unlikely to remain the defining story even a few quarters from now. We are on the cusp of something far greater. The sooner we recognize this, the better prepared we’ll be for a future shaped not by chatterboxes or agents, but by minds we are only just beginning to comprehend.

The author is a Singapore-based innovation investor for GenInnov Pte Ltd.