The strange dichotomy of AI

Like everyone else, I've been using AI for a while now and it's been a strange experience.

I can't deny that it's a disruptive technology and that it's quite useful, but the more I use it, the more I feel that there are things that are highly dangerous, wonderful, and strange, so I will try to put my thoughts here and update them as things evolve.

It saves time?

It's undeniable that it saves a lot of time, when used correctly. Going through dense documents to get the highlights, or pulling meeting notes and per-person tasks automatically, is a great time saver. No longer do we need to take notes and write follow-up documents (I did a lot of this in my previous life to keep people organized and not let them forget about the tasks that they needed to do), or spend time writing stuff that no one reads because we can now have AI doing that job for us (slightly joking).

I said "used correctly" because I also see it being used for the sake of being used. For example Meta created leaderboards to see who used more tokens as well as Amazon to incentivize people to use it, just to backfire when people start gaming it and generate a lot of useless tasks just to be ranked higher (and not get fired).

Using AI for the sake of "saving time" and "improving productivity" is absurd. It's like having leaderboards for who runs the most Google searches to "save more time". It's only useful if you make it useful.

AI-first is not the way to do it

Microsoft is pushing more and more AI everywhere, to a fault in my opinion. You have an AI button everywhere that you can access, especially in the Office portal where previously you had a nice list of things that you can access and now you have just a "ChatGPT"-like UI. What if I want to change something on my site, add a new user? Do I need to use the chat all the time to do these things?

I've lost track of the times that I wanted to access the admin portal and needed to go and search for the URL up until I needed to create a bookmark. Previously it was here for me to click.

AI is wonderful for some tasks but not all and having this shoved into your face is a bad decision, even if it was the best, super smart, and always solved your problems quickly (it isn't). If you don't agree with me try developing a Power Automate Flow and then ask Copilot to explain an error and see what you get.

No one likes to be on the other side

But the thing that I find particularly interesting is that everyone is using it, but everyone hates being on the "other side". Microsoft and Google happily show you AI-generated email responses that you can quickly send. It saves you time. But it sucks for the other person who receives them if they realize it was generated by AI. Even if you use this all the time, if you receive a reply or a report that was generated by AI you would be annoyed or even discard it altogether. So people are happy to "save time" and use AI but angry when they are "on the other side".

So if AI is a miracle why does it only work on one side and not both? Why aren't people happy when things are not written by humans but happy when they can "delegate" their work to an AI agent?

AI replacing people

So you're an amazing manager that uses AI so efficiently that you do all your job using AI. Good for you, you save a lot of time. But now the question is, "what is your added value", or more bluntly "why do we need you?".

This is one of the things that companies are evaluating and clearing up massive parts of the companies, especially middle management. Don't get me wrong, I've seen a lot of companies where some managers were only useful as glorified mail router systems where they only provided information from one layer of the company to the other, but having it replace people just for the sake of "AI powering the company" will have dangerous results.

If you don't believe me, let's check Meta again. Instagram was "hacked" recently because their support was AI-based. So if you provided fake information the system happily restored the account thinking it would be yours. Humans are fallible as well, but replacing humans with AI doesn't generate better results. Companies are already hiring people again that they fired when they tried to follow the AI hype machine.

So if AI is such a good thing, why is Anthropic, a company famous for saying developers will be obsolete in 6 months, hiring more and more developers and doesn't allow them to use AI on interviews?

Think!

I wrote a while ago about "With AI You need to know more not less" and I 100% stand by it. When I use AI heavily for my tasks I don't feel more productive. I feel more tired. Managing all these "agents" in parallel, letting them think and guiding them through what I want done, is super tiring. Probably I'm showing my age here, but it's not productive for me. I lose control of things and then problems show up. Bugs in code, things that are inconsistent and worst of all I don't think. I let agents think for me and that's not good.

Thinking is being devalued a lot. Companies say it's a good thing. It's not.

I saw this a while ago when Google appeared. People complained that going to Stack Overflow and getting an answer (copy and paste a random piece of code) was easier than coming up with one ourselves, but it's different. We still needed to work on the problem and getting an answer was a shortcut not a replacement.

AI is trained on our collective reasoning and thinking so if we delegate thinking to AI what's left? I'm not even talking about high level thinking, but simple things like writing emails or meeting notes were good ways for us to organize ideas in our minds and, sometimes, find things that didn't make sense. Now we ask AI to "check the meeting notes and see what I need to do", meaning that we're training our brain not to pay attention to anything. Our attention span is narrowing towards zero, and with it will come a point where we can't sit 5 minutes and think about something without asking AI. It's easier, but it doesn't mean it's better.

The Atlantic has a nice article about "The Film Students Who Can No Longer Sit Through Films" and soon we will have "Developers who don't know how to code" or "managers who don't know how to analyze a report".

Final Thoughts

I've built this site to provide answers to problems, so I understand that this article doesn't provide any solution whatsoever, quite the contrary. It only raises questions. That's the objective. AI is a wonderful disruptive technology and, like all the other technologies, will take a bit of time to stabilize and find its place in society. It happened with all technologies like the internet, for example, and now you don't see anyone saying that we should go back to a time without it.

We need to give it time, but also don't lose the things that are important, like thinking, or using it for what it's good for, not for everything.

Photo by Dan Meyers on Unsplash