Is AI ready to write our laws and make our home-tech?

Recently, Estonia launched an AI system to review parliamentary bills after a drafting error cost the government millions in tax revenue. Similar moves are happening in music production, where AI chatbots now accept complex instructions to manipulate audio files, and in robotics, where new home companions use human-like hands for daily tasks. Meanwhile, entertainment platforms are releasing interactive, AI-animated shows that allow viewers to chat with characters once an episode ends.

These new capabilities rely on a shift from text-only systems to what developers call agentic systems. A standard AI model usually acts like a librarian, retrieving knowledge it learned during its initial training period. An agentic system, however, has a set of digital hands. It connects to external tools, such as software for editing music, document databases for legal review, or the electrical motors in a robot’s fingers. These systems process a goal from a user, break it into manageable steps, and execute those steps one by one using specific software commands or physical hardware triggers. The accuracy of these actions depends on the software having a clear rulebook to follow, which is why Estonia argues that automated decisions work best for clear-cut rules, while human judgment is required for complex, nuanced situations.

The core danger is not that AI will take over, but that it will act at scale without a clear chain of responsibility. As these systems move into our homes—controlling lights or laundry—and our governments—drafting the rules we live by—the human oversight layer must remain robust. The challenge is ensuring that when an AI makes a mistake, whether it is a tax loophole or a robotic mishap in your living room, there is still a person accountable for the outcome. Trusting AI to assist is one thing; delegating authority to it is an entirely different risk that requires us to rethink how we define accountability in an automated world.