RIYADH: Far from fearing a future powered by AI, researchers at King Abdulah University for Science and Technology are using it to uncover long-hidden secrets about Saudi Arabia’s past.
Prof. Bernard Ghanem, a specialist in computer vision and machine learning, said that in particular, AI is helping to discover archaeological sites that have yet to be unearthed.
“AI has applications in every part of our lives: analyzing the present, the future as well as the past,” Ghanem told Arab News.
His team at KAUST has trained AI models, using satellite data and images of known historical sites, to assist them in the identification of undiscovered sites across the country, he said. The resultant findings have fueled further archaeological research and are helping to preserve the Kingdom's rich cultural heritage.
However, archaeology is just one of the many areas of study in which Ghanem’s team is exploring the potential benefits of AI technology.
At the Image and Video Understanding Lab, for example, researchers are focusing on four main applications of AI, mostly rooted in machine learning, a branch of AI in which systems use existing data to help them solve problems using statistics and algorithms.
The first involves building machine-learning models specifically for use with video to harness the popularity and power of streaming.
“Video is the biggest big data out there; more than 80 percent of the internet traffic that we see is because of video,” said Ghanem, whose team is developing tools to analyze, retrieve, and even create videos, thereby leveraging the ubiquity of video in new AI applications.
The second application, which uses machine learning and deep learning to aid automation, is investigating the ways in which two-dimensional simulation data can be translated into the 3D world, with potential applications in gaming, robotics and other real-world scenarios.
“How do you, for example, play a game in the simulated world and then have that … work in the real world?” Ghanem said.
The third is exploring the foundations of machine learning, with a focus on identifying weaknesses in generative AI models and finding ways to improve them and prevent failures.
Ghanem compared this process to building immunity, whereby the AI models are deliberately “broken” to help understand vulnerabilities so that can be addressed and the models strengthened.
The fourth application involved the use of AI for science, specifically its use in efforts to advance chemical research.
Ghanem said his team is developing AI models able to act as virtual chemistry assistants by predicting the properties of molecules and perhaps discovering new compounds. Such innovations, he added, could play a critical role in the study and research of topics such as catalysis and direct air capture, thereby boosting efforts to combat climate change.
Ghanem also highlighted the environmental potential of AI, and the new Center of Excellence for Generative AI at KAUST, which he chairs. The center, which is due to open on Sunday, will explore four key pillars of research relating to: health and wellness; sustainability; energy and industrial leadership; and future economies.
“That’s where we’re going to focus on GenAI methods for sustainability,” Ghanem said.
