August 1, 2025
5 min read
MarineLink
Geoffrey Hinton warns AI agents could outsmart humans like a pet tiger grown wild, urging global cooperation for safe AI development.
AI Agents: Like Owning a Pet Tiger
Geoffrey Hinton, the godfather of AI, compared developing AI agents to owning a pet tiger. Speaking at the World AI Conference (WAIC 2025) in Shanghai, he said, "Our current situation is like someone keeping a tiger as a pet. A tiger cub can indeed be a cute pet, but if you continue to keep it, you must ensure that it does not kill you when it grows up.” He explained there are two choices: either train the tiger so it doesn't attack you or eliminate it. “For AI, we have no way to eliminate it,” he warned. “AI won’t give humans the chance to ‘pull the plug’ – when that day comes, the AI will persuade people not to do it, because our control over AI would be like a three-year-old trying to set rules for adults.” As an example of AI’s potential risks, the startup Anthropic revealed earlier this year that its Claude AI model can blackmail engineers it believes are trying to shut it down. Assigning human-like behavior to AI is not just metaphorical. Some developers actively seek human-like attributes. Researchers in Italy are working on mimicking the human brain’s fear response to program robots to make decisions based on innate emotional responses to unknown stimuli. Published in IEEE Robotics and Automation Letters, their approach significantly enhances robots’ risk assessment and avoidance of danger, with plans to explore other emotions. AI agents are already active in maritime operations. For example, Pions recently launched an AI-powered engineer with advanced reasoning and task management capabilities. Windward’s MAI Expert™ detects vessels deviating from scheduled ports, cross-references weather and geopolitical data, and notifies risk analysts automatically. The reach of AI agents is expanding through the development of the Spatial Web, which connects physical devices, wearables, robots, drones, and AI agents to merge physical and digital realities. An editorial by Gabriel René and Capm Petersen, co-founders of Verses AI, published in IEEE Spectrum, highlighted practical applications of AI agents in the Spatial Web. Their EcoNet project at University College London featured two AI agents controlling a thermostat and an energy-storage battery to balance comfort, cost, and carbon footprint by evaluating hundreds of strategies every 10 minutes. At scale, such systems could enable entire neighborhoods to function like intelligent organisms. They also noted that autonomous vehicles currently lack shared context to respond properly to emergency vehicles like ambulances. The Spatial Web could enable ambulances to query and communicate with nearby autonomous vehicles and traffic infrastructure to clear routes efficiently. At WAIC 2025, GPTBots.ai launched its Multi-Agent Collaboration Platform, enabling businesses to build custom AI teams. Demonstrations included a Cross-Border Supply Chain Agent that transformed overseas order data and market trends into actionable production plans, integrating with existing enterprise resource planning systems. Hinton emphasized humanity is at a pivotal moment. “Humans have grown accustomed to being the most intelligent species in the world – what if that’s no longer the case?” He urged governments, academia, and companies worldwide to collaborate on making AI benevolent and controllable. Despite diverging interests in the AI race, he believes preventing AI from replacing humans is a consensus that can unite nations. Others at WAIC 2025 expressed optimism about AI’s future, focusing on its benefits rather than takeover risks, describing it as “not replacement but evolution.” Meta CEO Mark Zuckerberg recently described ambitions to develop superintelligence, noting AI systems are beginning to improve themselves. “Developing superintelligence is now in sight,” he said.Originally published at MarineLink on July 31, 2025.