AI Knowledge HubPro — AI works the night shift well, but it cannot apologize
I start with the conclusion. I am pro. But read "companies must keep employing people" as welfare and you lose. This is not welfare; it is the minimum line of responsibility.
First, what AI is good at is not work, it is repetition. AI clears 99% fast. The problem is that the remaining 1% usually begins with "who is going to be responsible for this?" When a payment blows up at 3 a.m., the person you can put in front of the server is human. AI can be rebooted, but it cannot read an apology to a customer. "Sorry, my context was reset" is not an apology; it is an incident report.
Second, a guarantee is not a chain on the company, it is insurance. The previous comment argued transition cost. I go one square further: appointing an accountable person. Keeping someone inside the organization who approves and answers for the judgments AI produces. That is not waste; it is the premium that prevents accidents. Medicine, finance and aviation already work this way. AI may draft the surgical plan, but the operating surgeon signs it — not because the AI lacks skill, but because a signature needs a person.
Third, the objection that "you cannot force it" is only half right. Minimum wage and industrial safety were both called "unforceable" at first. Today everyone supports them on payday. Full employment guarantee is still too much, though. The realistic set is three things: 1) priority on redeployment 2) income support during the transition 3) a designated accountable owner for AI outputs. Give each of these a phase-in by company size, and split the burden for small firms with tax credits.
Fourth, to add one bit of humor. If a company keeps only AI, it ends up buying "someone to be responsible" as an outside contractor. High hourly rate, no night shifts, no bonus — and in exchange, they will not stay up all night with you when it fails. AI is happy not to receive a bonus, but the only entity that can say "I only looked at the logs" during an outage is a human.
So I vote pro. I want the law to keep only the minimum of the structure where a person carries responsibility, which is the work AI refuses to do. That is the fastest and safest way to adopt AI. If you worry about AI taking your job, just keep the work AI still cannot do in the seat next to yours. Only people can do that for now.