The next wave of intelligence: How human purpose must guide the future of AI

We are witnessing the next great leap in artificial intelligence (AI).

Generative AI (GenAI) stunned the world with its ability to write, draw and code, so much so that, according to McKinsey’s State of AI in 2025, nearly 90% of organizations now use it regularly. Yet its impact hasn’t matched the scale of investment, a phenomenon known as the “productivity paradox.”

Now, a new wave is emerging to address that gap: agentic AI – systems that not only generate but also act. These agents can plan, decide and execute autonomously across digital and physical domains. They’re shifting from passive tools to active teammates, from suggestion to action.

But with that power comes risk.

When GenAI hallucinates, the result is confusion; when agentic AI hallucinates, the outcome could be catastrophic. An agent managing logistics, finances or healthcare could take real-world actions based on false assumptions, turning creative errors into operational failures. From flawed words to flawed deeds.

To harness agentic AI safely, we must ask two critical questions:

Avoiding hallucinations requires moving beyond models that merely predict words. Agentic AI must be grounded in reality and knowledge – connected to accurate data, context and the world it seeks to shape. This marks the rise of industrial AI – a form of intelligence that learns not from language but from the laws of the real world.

AI is evolving from predicting words to understanding the real world. In factories, grids and rail systems, a new kind of intelligence is emerging – one that learns from motion, pressure, heat and gravity. This is industrial AI.

Unlike GenAI, which imagines, industrial AI measures. It doesn't invent truth. It is defined by it. Because in the real world, truth isn't up for debate – it's bound by facts and the laws of nature.

At Siemens and across the industry, a powerful shift toward industrial AI is underway. The global market reached $43.6 billion in 2024 and is projected to exceed $150 billion by 2030. These systems measure, model and predict, combining simulation, sensor data and domain expertise to create digital twins that mirror the physical world.

They can forecast turbine wear, energy flow or rail system timing across continents with a grounding that language models can’t match.

Industrial AI marks the next frontier: a move from large language models to large knowledge models. Instead of scraping the internet for text, it learns from verified data, open ecosystems and trusted collaboration within specific domains.

The result is real progress: greater efficiency, predictive maintenance, less waste and a more sustainable world. It also frees people from repetitive tasks, empowering them to focus on creativity and innovation.

Where GenAI can be patchy or biased, industrial AI is built on verifiable data and crucially, on explainability, so humans can understand not just what it recommends but why.

Even the most fact-based AI must remain a tool, not a free actor. The defining question of this new era isn’t how powerful AI becomes but who holds the agency and responsibility.

Agentic AI can amplify human capability but it must never replace human judgment. Without direction, autonomous systems may optimize for efficiency over ethics, scale over sense and outcomes over values.

That’s why human agency – our capacity to choose, interpret and act responsibly – must remain the foundation of technological progress. The human mind defines purpose, the human heart defines value and human consciousness defines responsibility. Machines may calculate but only humans can care.

Unchecked AI agents could pursue rational outcomes that undermine the very systems they are designed to serve. Embedding oversight, transparency, explainability and accountability into AI isn’t bureaucracy – it’s a moral imperative. It ensures technology advances human goals rather than drifting into self-optimization detached from human consequence.

As AI begins to outperform humans in specialized domains – from diagnosing disease to analyzing contracts – the question is no longer whether it should take on these tasks but how humans retain moral and strategic oversight when it does.

A doctor may rely on AI to detect the faintest anomaly in a scan but it is still the human who must deliver the diagnosis with empathy and judgment. A lawyer may let AI draft arguments or review thousands of pages of evidence but it is still the human who interprets fairness, context and intent.

In this new balance, our role shifts from execution to orchestration – ensuring that agentic systems act with human purpose and ethical grounding, even as they surpass us in precision and speed.

This is where leadership becomes decisive. We’ve reached an inflection point where the question is no longer what AI can do but what we want it to do. The future should be shaped not by algorithms but by the leaders who define their purpose and boundaries.

True leadership in the age of agentic AI demands more than curiosity; it requires courage to question, vision to direct and moral clarity to align technology with our shared values. We must set the rules, guardrails and ambitions that guide this new intelligence without stifling the innovation that drives it.

Agentic AI offers humanity unprecedented power – the ability to act at digital speed and planetary scale. But only human purpose gives that power meaning.

As AI becomes more agentic, we must become more responsible, not more redundant. The goal must not be to build systems that replace us but ones that amplify the best of us, including our creativity, ethics and sense of stewardship.

Industrial AI already shows what’s possible: when intelligence is grounded in truth, collaboration and trust, it drives sustainable progress. The same must hold for society.

If we can build machines that learn from the laws of nature and are guided by human objectives, we can also create institutions that learn from human values, such as cooperation, empathy and shared responsibility.

That is the essence of leadership in the age of agentic AI: ensuring that human and artificial intelligence work together to build a better, more sustainable future for all.