At Saudi Arabia’s Leap 2025 event in February, the kingdom rolled out a major investment in artificial intelligence (AI). The conference buzzed with pivotal discussions around AI, such as how to extract business value from it, the hurdles of developing agentic AI, and what the future holds for robotics and causal AI.
Yaser Al-Onaizan, CEO of the National Center for AI, zeroed in on agentic AI, which he described as AI that acts on our behalf. He pointed out that while large language models (LLMs) grasp how language works and can provide information, the true promise of AI lies in its seamless integration into our daily lives. “It should be invisible, listening, understanding, and taking action on our preferences without us having to ask,” Al-Onaizan said. Imagine interacting with an AI model that not only tells you about flights but also books them or secures a hotel for you. But for AI to responsibly operate in that capacity, it must understand “common sense” to make decisions that align with human needs.
Lamia Youseff, founder of Jazz Computing, shared her insights on AI’s evolution. She outlined key phases:
- Enterprise AI – This phase harnessed big data for analysis, leading to a wave of new products and services.
- Agentic AI – Expected to roll out in the next couple of years, this involves agents taking commands and executing complex actions.
- Robotics and humanoids – These require advanced communication skills and a deep understanding of human language, particularly for applications like driverless cars.
- Causal AI – This will push boundaries by predicting intricate real-world events, including fluctuations in the stock market.
Aiden Gomez, CEO of Cohere, highlighted how to extract value from AI for businesses. “You need technical know-how to create useful applications from LLMs,” he remarked. But businesses face challenges, like integrating AI with existing systems while maintaining security. “Companies need solutions that can integrate effortlessly,” Gomez emphasized. As AI moves from proof-of-concept to production, data sensitivity rises, making security a top priority.
He noted that agentic AI hasn’t yet mastered reasoning. “Right now, a model takes the same time to answer basic math as it does to prove complex theorems. That’s not efficient,” he explained. Improving reasoning and enabling models to learn from user interactions will transform agentic AI significantly.
Lambert Hogenhout from the United Nations raised important concerns about AI’s impact on society. He highlighted the risks of losing human autonomy, identity, and purpose, especially if AI starts replacing jobs. He cautioned against blind acceptance of AI’s capabilities, warning that it could undermine what makes us uniquely human, including creativity and connection.
“Right now, we’re too passive. We remain reactive to new AI developments without anticipating their effects on our lives,” Hogenhout stated. As we increasingly rely on AI for communication, he questioned the necessity of our individuality if AI can replicate our personas. He pointed to apps that aim to keep the memory of loved ones alive, illustrating the potential consequences for identity.
Hogenhout stressed the need to prioritize meaningful connections and fulfillment in our society. “As jobs evolve or disappear, we must harness our unique skills, especially our ability to cooperate and innovate. Otherwise, we risk alienating ourselves and creating societal fractures,” he warned.