By 2035, “who is responsible when an algorithm decides?” will be settled mainly in courts and councils, not seminars

Prediction: by 2035, the practical question of who is responsible when an automated system makes a harmful decision will be answered mainly through court rulings, regulations and procurement rules, with explicit duties for the people who build, buy and deploy the systems, rather than being left as a topic for philosophers.

Reasoning: philosophers have long discussed machine ethics and responsibility, including what we can reasonably hold a designer, an operator or a user accountable for. But real disputes, over lost benefits, wrongful decisions or accidents, force institutions to decide. Legal systems usually settle such questions case by case, and then codify them.

Confidence: moderate, about 60 percent. The details will vary between countries, and some questions will stay open.

What would prove us wrong: if, by 2035, there is still no widely applied rule about who is accountable for automated decisions in major areas such as credit, hiring and benefits.

How to check: review legislation and notable court rulings in 2026 and 2035 for clear allocation of responsibility.