ISLAMABAD: A machine can solve mathematics problems that challenge some of the world’s brightest minds, write sophisticated computer code and answer scientific questions that once required years of specialist education.
Yet the same technology can misread an ordinary clock, invent information or struggle with a task that a child might complete without difficulty.
This contradiction lies at the centre of one of the most important scientific questions of the modern era: how close is artificial intelligence to becoming smarter than humans?
By 2026, AI has moved far beyond the experimental systems that once struggled with basic conversations. Advanced models can analyse documents, generate realistic videos, assist scientific research and carry out increasingly complicated computer-based tasks.
But whether these achievements amount to intelligence comparable to the human mind remains an open question.
Surpassing Humans
The speed of AI development has surprised researchers and policymakers alike.
Stanford University’s 2026 AI Index Report found that leading AI systems now meet or exceed human performance benchmarks in several demanding areas, including advanced scientific questions, mathematical reasoning and coding.
In 2025, AI systems developed by Google DeepMind and OpenAI achieved gold-medal-level results on problems from the International Mathematical Olympiad.
Such achievements would have appeared extraordinary only a few years earlier.
AI is also transforming professional work. Software developers use it to write and review code, researchers employ it to examine complex information, and businesses increasingly rely on automated systems for analysis and customer services.
These developments suggest that machines are already demonstrating abilities beyond those of most humans in certain specialised tasks.
However, performing exceptionally well in a particular field is not the same as possessing broad human-level intelligence.
Intelligence and Understanding
Human intelligence is not measured solely by the ability to answer difficult questions.
People learn from experience, adapt to unfamiliar situations, understand social relationships and make decisions in environments where information is incomplete.
Modern AI systems operate differently.
They learn patterns from vast quantities of data and use those patterns to generate responses, predictions or actions.
Although their performance can resemble human reasoning, researchers continue to debate what these systems genuinely understand and how their abilities compare with human cognition.
The 2026 International AI Safety Report describes the capabilities of advanced AI as uneven.
ALSO READ: World Bank Plans to Boost Digital Connectivity in Developing Countries
A system may perform exceptionally well on a difficult examination while failing at a comparatively simple task involving physical space, unfamiliar instructions or several consecutive decisions.
Stanford’s AI Index offers a striking illustration: despite major advances in mathematical reasoning, a leading model correctly interpreted analogue clocks only about half the time in one evaluation.
Such inconsistencies show why success on individual benchmarks cannot establish that AI has achieved general human-level intelligence.
Artificial General Intelligence?
The next major milestone in AI development is often described as artificial general intelligence, or AGI.
Unlike systems designed to perform particular tasks, AGI generally refers to AI capable of handling a broad range of intellectual activities with flexibility and competence comparable to humans.
However, there is no universally accepted definition or agreed test for determining when AGI has been achieved.
Some researchers emphasise performance across many cognitive tasks. Others argue that genuine general intelligence must include the ability to learn independently, respond reliably to unfamiliar situations and complete complicated objectives without constant human supervision.
This disagreement makes predictions particularly difficult.
A model might outperform people in mathematics, programming and scientific examinations while remaining unreliable in practical situations requiring judgement, common sense or physical interaction.
For that reason, claims that AI has already reached human-level intelligence require careful scrutiny.
Smarter Than Humans?
The possibility of artificial superintelligence raises an even larger question.
Superintelligence is commonly understood as a hypothetical form of AI that would substantially exceed human cognitive abilities across a wide range of fields.
If developed, such systems could potentially accelerate scientific discoveries, improve medical research and help address problems that currently require enormous human effort.
But there is no established timetable for their arrival.
The International AI Safety Report notes that projections for AI development through 2030 vary considerably.
Some scenarios envisage rapid advances, while others anticipate slower progress or substantial limitations.
Researchers have also begun measuring AI progress through its ability to complete longer tasks independently.
The research organisation METR reported that the length of software-related tasks frontier AI systems could complete at a specified success rate had historically doubled approximately every seven months.
That finding points to substantial progress in autonomous capabilities.
However, METR cautions that its measurements focus heavily on software and related technical tasks. They cannot establish when AI will be capable of performing every human job or achieving general intelligence.
The Risks of Machines
As AI becomes more powerful, concerns are growing about whether society can manage the technology safely.
Advanced systems can already produce convincing misinformation, assist cybercriminals and generate inaccurate material that appears authoritative.
Increasing autonomy could create additional challenges if systems are permitted to make consequential decisions or operate digital tools without adequate oversight.
In July 2026, a United Nations scientific panel warned that AI development was advancing faster than researchers’ understanding of its potential risks and governments’ ability to regulate it.
The panel highlighted concerns involving deceptive behaviour, autonomous systems and the possibility of losing effective control over increasingly capable technologies.
These warnings do not establish that AI will become uncontrollable.
They underline the need for independent testing, transparent safety evaluations and safeguards proportionate to the capabilities of the systems being deployed.
The debate also extends to employment, education, privacy and economic inequality.
If AI can perform a growing share of intellectual work, governments and businesses will face difficult questions about how to distribute its benefits and support workers whose responsibilities change.
Human Intelligence?
The rise of AI does not necessarily mean human intelligence will become irrelevant.
Machines may eventually outperform people in many forms of analysis, calculation and information processing.
But human societies still depend on accountability, ethical judgement, relationships and decisions about which goals are worth pursuing.
A system capable of identifying an efficient solution cannot independently determine whether that solution serves the public interest.
Those questions require human institutions, values and oversight.
The challenge, therefore, is not simply whether AI can become more intelligent than humans.
It is whether societies can develop the knowledge, regulations and institutions necessary to use increasingly capable systems responsibly.
Science Fiction
Artificial intelligence has already crossed thresholds that once seemed distant.
It can solve advanced mathematical problems, support scientific investigations and complete complex digital tasks at a level that would have been difficult to imagine a decade ago.
Yet its weaknesses remain substantial, and researchers have not established that today’s systems possess the broad, reliable intelligence associated with human cognition.
Whether AGI arrives within a few years, takes decades or requires fundamentally different technological approaches remains uncertain.
What is clear is that the consequences of AI are no longer confined to predictions about the future.
They are already affecting how people work, learn, create and make decisions.
The defining question may ultimately be less about the moment machines become smarter than humans and more about whether humanity is prepared for a world in which machines can perform an expanding range of intellectual tasks.
