Or why AI agents are here, but we should avoid it in everyday life.

I started to hate it, but since AI has been one of the most popular topic in my buble and I’m pretty close to it - it feels right to keep talking about it. Because if engineers like me won’t - we might end up in a messy world the other day.

A something-bot has changed it name 2 times within a week and captured attention of thousands of people. People are giving away accesses to their Internet identities like these are candies during the Halloween. E-mails, calendars, notes. What happens if some illegal information will be found in these notes, how are you gonna protect yourself? What happens when all the password to your accounts will be changed - e-mail is still a basic, first-level communication channel over the Internet. What if your reputation will be destroyed by fake appointments created by AI?

These are edge cases, but absolutely real risks that can happen when access are granted to any 3rd party. But there is one thing that irritates me more: what are the tasks people are giving to the AI? „Cleanup my inbox” - was loud example. What is the problem there? Most of my inboxes consist of rare e-mails from friends or family, and most if these are invoices, password restorations or account setups. And the biggest part is spam. Sales, new product announcements, „we got this for you” and other attempts to make me consume more and spend more. If cleaning that mess is the hard tasks we need AI for - I’m confused about the world we’re living in.

I intentionally avoided professional e-mail, calendar and other notes. Because there is no way AI is getting access there. Because what might be the reason? If there is something that requires my decision - it requires MY decision, not an AI decision. Being in a leadership position I’m using AI to explain technical vision. But I refrain from using it when speaking to people about what’s human. Feelings, hard situations, problems. People need support from me, not from algorithms. Means it can’t replace me in that process as well.

Pull requests and coding? I don’t have an issue to be specific and opening dedicated console tab, and asking specific AI to do specific thing. Orchestrating is not the hardest part. And if you dream of „AI-agents based team” - I have an upsetting news for you. You will literally die because of the amount of changes and responsibility, if you’ll try to spin it up on the large scale.

Why people sign contracts before they start working? Why companies outsource software, hardware, services? It’s not only getting the result: accountancy software, properly working ACs and clear floors. It is responsibility shift. Someone certified software to meet all the legal requirements. Someone else studied and certified to properly install ACs, so entire building won’t get on fire. And someone takes care of people to be on time to wipe the floor.

How are you going to shift responsibility to AI? Who’s actually taking care of meeting the goals and delivering the results? If there’s no human with real decisions and responsibolity - that’s like living in an imaginary world. A beautiful castle might break down in seconds. Burying you under the paper bricks it was built of. A ton of paper is still a ton. That’t the pressure we can’t survive. Please don’t get fooled by promises.