You bring up valid concerns, but I believe we can do more than simply banning AI for non-researchers and the military. We need to generate numerous ideas to address these issues, discuss them, and promote the best solutions using all available resources.

1. Addressing unemployment:
   We could consider grading AI systems based on approved tests. Basic systems should be accessible to everyone, while more advanced systems should serve as assistants to intellectual workers such as programmers, writers, and scientists, rather than replacing them. If a company replaces an employee with AI, they should contribute additional taxes to a Universal Basic Income (UBI) fund. People do not need money for sake of money. They need food, clothing, shelter, healthcare, and entertainment. AI could make these essentials much more affordable and accessible. Governments should support citizens during the transitional period, just as they did during the pandemic. This transition will be gradual, allowing for improved efficiency, lower prices, and increased production of goods and services. If necessary, people could receive support in the form of vouchers for essential items.
2. Combatting intellectual atrophy:
   A similar situation arose when people transitioned from physical jobs to office work, leading to the rise of gyms. As parents are legally responsible for their children go to school, we need to create mental exercises, potentially making them mandatory. AI could help tailor these exercises to individuals, making them efficient and virtually cost-free. These activities could take the form of engaging games. One form of motivation could be government assistance linked to individuals regularly participating in such exercises or passing intellectual tests, provided they already have intellectual jobs or hobbies.
3. The impact on democracy:
   Large-scale AI has the potential to integrate people's opinions on various levels (local/state/federal/world) more effectively than politicians. Many government positions could be replaced by AI, which would not be susceptible to bribery. Remaining roles would be monitored by AI, allowing any person to access clear and easy-to-understand explanations of a politician's positive or negative actions. This presents an opportunity for positive changes in democracy.

Why is that? To avoid surprises it would be enough to announce that an AI boy of 14 years old is always available to answer your questions. He won't answer all your questions, just the ones he likes. If you don't like that boy, don't ask him anything. Many people, especially young people, will still find this bot useful and fun.
We're gonna have many, many bots with different personalities, so everyone can find their favorite.