Artificial intelligence is no longer a technology of the future. It is already reshaping medicine, education, research, communication, business and production—and doing so at remarkable speed. Stanford’s AI Index reports that 88 percent of surveyed organisations used AI in at least one business function in 2025, up from 78 percent in 2024. Generative AI use also rose from 71 to 79 percent over the same period. AI, in other words, is not only becoming more capable; it is becoming more pervasive.
But one fundamental question cannot be avoided: as AI’s capabilities and use expand, are our institutions and our ability to control its risks advancing at the same pace?
The risks are no longer hypothetical. Misinformation and deepfakes can distort public understanding; privacy violations and algorithmic bias can undermine rights; and AI may reshape employment while encouraging excessive reliance on machine-generated decisions. The number of documented AI-related incidents also rose from 233 in 2024 to 362 in 2025.
These concerns gain added weight from warnings by AI pioneer Geoffrey Hinton, who has raised the prospect of widespread misinformation, major changes to employment and the possibility of AI eventually becoming more capable than humans. More than 700 AI researchers and prominent figures have likewise called for reducing the risk of extinction from advanced AI to be treated as a global priority.
AI researcher Connor Leahy points to another, less visible problem. We know how to build and train AI models, yet we cannot always fully explain how complex decisions emerge within them. He has described this gap in understanding as an “unsolved scientific problem”.
That makes the issue larger than what AI can do. It is also about how well humans can understand, supervise and control increasingly powerful systems. There is still no single, binding international safety framework for frontier AI. If highly capable systems fall under the control of actors without sufficient accountability, the risks of misuse could grow.
AI also carries an environmental cost. The IEA projects that global data-centre electricity consumption could nearly double to around 950 terawatt-hours by 2030.
The answer is not to halt AI’s development, but to advance safety, transparency and accountability alongside it—through rigorous standards, independent testing, human oversight, international cooperation and worker retraining.
The final question, therefore, is as much about human intelligence as artificial intelligence: as AI grows more intelligent, is human responsibility growing with it? Can we harness artificial intelligence for human welfare while keeping its risks—and our own security—firmly in view? Ultimately, achieving that balance may be the clearest measure of human intelligence.
By Salam Abu
Editor
