- AI tools herald a boundless scientific revolution in accelerating treatments and medicines that can transform human lives
- AI “hallucinations” pose a new global challenge related to misinformation and the lack of full reliability
- AI systems face enormous challenges in electricity consumption, and comparisons with the human brain remain unfair due to the vast gap in energy efficiency
A number of Nobel Prize laureates and recipients of prestigious scientific awards affirmed that artificial intelligence, despite its rapid development, remains at an early stage when it comes to achieving major scientific breakthroughs.
They stressed that the decisive and leading role in discovery pathways still lies with the human mind, with its capacity for experience, creativity, interpretation, and analysis.
This came during the forum titled “Artificial Intelligence Sciences: Is AI Capable of Discovery?”, held as part of the World Scientists Summit, which launched today alongside the World Governments Summit.
The World Scientists Summit is the largest global gathering of its kind, bringing together over three days more than 100 scientists and participants, including Nobel Prize laureates, recipients of other major international scientific awards, and leaders of research institutions. This takes place alongside the World Governments Summit 2026, scheduled from February 3 to 5, with February 3 designated as a joint day bringing together scientists with heads of state, governments, ministers, and leaders of international organizations and institutions participating in the summit.
The forum featured the participation of Professor Tony F. Chan, former President of King Abdullah University of Science and Technology (KAUST); Professor Yurii Nesterov, recipient of the 2023 Prize of the Association of Global Prize Winners in Computer Science or Mathematics and Research Professor at Corvinus University of Budapest; Professor Omar Yaghi, Nobel Prize laureate in Chemistry (2025) and Professor of Chemistry at the University of California, Berkeley; Professor Jack Dongarra, Turing Award laureate (2021) and Distinguished Professor Emeritus at the University of Tennessee, Knoxville; Professor Robert Tarjan, Turing Award laureate (1986) and Distinguished Professor of Computer Science at Princeton University; and Professor Arieh Warshel, Nobel Prize laureate in Chemistry (2013) and Distinguished Professor of Chemistry at the University of Southern California.
Participants emphasized that artificial intelligence is a powerful tool for supporting scientific research—such as accelerating high-performance computing, discovering new chemical models, and simulating complex processes—but remains limited in its ability to independently develop pioneering ideas or theories.
They also highlighted challenges related to the accuracy of AI-generated information and its high energy consumption compared to the human brain.
They stressed that AI can serve as an effective assistant in scientific discovery and solving complex problems, but it continues to rely on human expertise to guide it toward genuine innovation and major breakthroughs.
Discovery and Innovation
In his opening remarks, Professor Tony F. Chan affirmed that artificial intelligence has a growing capacity to support discovery and innovation.
He pointed out that the Nobel Prize in Chemistry was recently awarded to scientists who developed the AI tool AlphaFold, which predicts protein structures, opening new horizons in scientific research and accelerating the discovery of drugs and treatments capable of transforming human lives.
He explained that AI-based technologies have become a fundamental component of modern scientific research across multiple fields, particularly in the advanced integration of physics, mathematics, computer science, and neuroscience.
However, he emphasized that this does not rise to the level of replacing scientists who fundamentally reshape scientific paradigms, questioning whether AI could independently discover general relativity or quantum mechanics.
High-Performance Computing
In a keynote address, Professor Jack Dongarra explained the relationship between high-performance computing and artificial intelligence, noting that his work in developing algorithms and software relies on both theory and experimentation.
He said:
“Today, we see AI dramatically accelerating computing. In weather forecasting models, AI can generate 10-day forecasts in seconds instead of hours, with greater accuracy than traditional numerical methods”.
A Critical Perspective
In another keynote, Professor Robert Tarjan presented a critical view of what he described as AI tools’ “hallucinations,” explaining that such systems may generate inaccurate or unreliable information.
He added:
“At one event, the organizers prepared a biography for me using ChatGPT and awarded me two prizes I never received”.
Tarjan then posed a broader question:
“Can AI discover problems on its own? Can it independently invent general relativity or quantum mechanics? I believe these questions remain open.”
Virtual Reality
In the same context, Professor Arieh Warshel shared his experiences in integrating physics-based simulations with artificial intelligence.
He explained that in his work on enzyme design through simulation, he initially failed to achieve satisfactory results.
Driven by frustration, he turned to AI, using what he described as maximum entropy, and was surprised by what he called a “clear achievement of AI,” which helped him uncover a striking correlation with enzymatic catalytic efficiency and enabled him to predict an enzyme faster than natural evolution.
However, he noted that in other cases, such as the study of heart disease, traditional physics-based approaches proved more successful than reliance on AI.
For his part, Professor Yurii Nesterov emphasized the need to create a “virtual reality” for artificial intelligence, stating:
“We are at the beginning of learning how to do this efficiently. AI needs two things: first, virtual reality—a massive database—and second, an AI unit capable of decision-making”.
He add:
“Humans receive their surrounding reality for free, whereas AI is extremely costly”.
He added that AI possesses significant creative power and can discover new relationships and structures, but its conclusions pertain only to a model of reality rather than reality itself.
He Said:
“If the model is correct, the result is logical; otherwise, it is nonsense” .
Solving Puzzles
Professor Omar Yaghi presented an applied demonstration on what he described as “AI chemistry,” calling it “astonishing”,He explained that AI has enabled the formation of new chemical structures and molecules leading to technologies capable of harvesting water from the air.
“This took me 35 years to achieve,” Yaghi said, “which makes the speed at which AI accelerates discovery truly remarkable to me.”
In a related session, Professor Tony F. Chan moderated a panel discussion on the role of AI in scientific discovery.
Dr. Jayant Haritsa, Senior Professor of Computer Science and Automation at the Indian Institute of Science, noted that AI can uncover hidden relationships between graph theory and machine learning, but that the original ideas in both fields are human creations.
Dr. Amir Gohari, recipient of a 2025 European Research Council Grant and Associate Professor of Computer Science at the University of Oxford, explained that what interests him most is the fundamental limits of what AI and machine learning can achieve.
He noted that some problems are algorithmically unsolvable and therefore beyond the reach of AI systems.
“We often see on platform X people posting an image or video and asking whether it is real or fake,” he said. “An AI system often provides inaccurate answers, which illustrates a potential boundary of AI capabilities.”
Meanwhile, Dr. Hesham Omran, recipient of the UNESCO–Fozan International Prize for the Promotion of Young Scientists in STEM (2023) and Associate Professor of Electronics and Communications Engineering at Ain Shams University in Egypt, stated:
“We all agree that artificial intelligence is a powerful tool that has transformed most aspects of our lives. It serves as an effective assistant for scientists, but we must distinguish between routine science and revolutionary science.”
He added that while AI is capable of solving puzzles within the prevailing framework, it is unable to think outside that framework.
“AI systems face enormous challenges in electricity consumption, to the extent that they place pressure on power grids. Therefore, comparing AI to the human brain remains unfair due to the vast difference in energy efficiency.”
