For the past three years, we have been in the AI era. In the beginning, it was hard to adopt AI, and only a small portion of people were using it, but now it is the biggest trend, and is present in almost all areas.
I remember when I was in college and we started using AI in our subjects. At the time, the answers were not very good and were followed by a lot of mistakes. However, my teachers were already predicting that this would be the "future of search".
I believe AI is really powerful, and also can improve your productivity, but, it can also be your downfall. Using AI in the way that I see many people using it can cause long-term damage that, unfortunately, they are not even aware of.
The cliché, but still relevant quote: "With great power comes great responsibility".
Learn by doing
How many times in your life you truly learned something just by listening to it? or even by reading it? or only by passively watching things get done?
My first question right here is: How much can we learn by passively watching AI do everything?
I think this question can be broken down into different levels.
If you have never done something and you ask AI to do everything for you, I really believe you cannot extract much from that experience. I'm going to use programming as an example: if you want to create something from scratch and you do not fully know how to do it, using AI to build everything for you is a very bad long-term choice.
All the learning that comes from making mistakes is gone, and you may start thinking that everything works like magic all the time, which is definitely not true.
When we delegate the possibility to learn something new to AI, we are also delegating mistakes, risks, fun, knowledge, confidence, mindset and growth.
You make your choices, and then your choices make you.
Curiosity and Critical thinking
Are you only interested in getting things done, or do you also care about how things get done? How do you know the way you did something was the best choice? How can you know different approaches?
Unfortunately, this topic is not discussed as much anymore. I think the fast-delivery industry is trying to shut this topic down, this is no longer how the world wants to work. I can see a tendency to get things done as quickly as possible, and this approach bypasses critical thinking. The end result? A long-term damage, as we discussed.
In the new AI world, we do have more power in our hands, but what the industry seems to care about most is velocity, and the number of things we can get done. At this pace, using AI, we are producing bad outputs, and the reason is easy to find: We don't want to think anymore.
But why would we think? We have a tool that builds everything we want, why would we care about that?
If you want to overcome this new tendency, here are a few things you can do:
- Learn how the sausage is made
Do not take every answer or output for granted. Learn the pros and cons of the choice and understand why that choice was made.
- Do not take the easy road
It is very easy to see things working without thinking about how they work, but don't do that, swim against the tide, learn the basics behind what you are building, try to go deeper, read books, take notes yourself.
Sometimes, it is better to do something manually that you have never done than to use AI and extract nothing from the experience you had.
When was the last time you did something by yourself? How long has it been since you wrote something yourself, really paying attention to every detail, instead of writing the minimum possible version and asking AI to improve it?
Developer vs User
This section is more related to the software engineering area.
As developers, we build software and solve things that do not necessarily involve code. At the same time, writing code, in most cases, is the biggest part of our job.
The question, "How do you spend most of your time, as a developer or as a user?", which I heard during a college seminar presentation, has stuck with me since then. I remember taking some actions in my career after hearing it, it made me realize that I wanted to invest more time thinking and building things as a producer, and not as a consumer of the technology.
When we delegate everything to AI, we put ourselves back into the consumer/user perspective again. Sometimes, we give up our greatest power.
I was making some commits using the github interface, and when I was creating a pull request, Copilot suggested a commit message. At first glance, I thought, "that's cool", but then I noticed that I hadn't really thought about what I had built. I was just clicking buttons automatically, without reflecting on my work.
The next time you are building something, ask yourself: "Am I being a developer in control and aware of what I'm building? or am I just a user consuming every output AI gives me?"
The cost of the HYPE
Every time I open my browser or linkedin, I see a lot of articles, courses, videos, and content about prompt engineering and how to get better at using AI.
I know it is good to learn the basics and understand how to get more assertive answers and outputs from AI. However, in my experience, the fundamentals and the basics related to your field are the most valuable.
In the software engineer context, there are topics that never get old, and this knowledge makes the biggest difference in your career. Learning about DNS, memory management, software architecture, code design, CI/CD, cloud computing, and so on will sharpen you more than any prompt you learn.
You don't need to ignore the hype, I would only suggest learning the basics before jumping into the hype.
The cost of AI in your company
Have you thought about the impact of AI in your company? What is the cost of coding faster using AI? Is there any impact on the teams using AI? What about the domain knowledge, is it still relevant? How do you measure code quality when using AI?
- Measuring code quality using AI
Well, that's a tricky one. AI has been trained on a large amount of code from many sources, and it is hard to define what good code and bad code really are.
Would you trust AI to decide whether code is good or not? and based on which concepts? I think most of the discussions in the software engineering area are about principles and concepts we have, and whether or not we should follow them.
AI can read entire codebases easily... Humans cannot.
- The impact on teams using AI
First, I would like to start talking about the domain knowledge. Companies used to care more about the domain knowledge, and from my experience, it plays a huge role in day-to-day operations, bug fixing, and generating value through new features.
If we start using AI to solve problems in day-to-day operations instead of training people to build strong domain knowledge, the long-term outcome can be harmful. What if you need to fix something urgent in production? Would you trust AI entirely to do it? How likely is it that AI will fully understand the context and provide the best solution for the scenario?
The combination of AI and someone with strong domain knowledge is powerful, but choosing only AI is not a wise decision.
Second, if we start isolating people to work only with AI, how will knowledge sharing happen? There are things that only day-to-day conversations or a pair programming session can provide. Also, if we isolate people to work only on AI, we lose interactions and, consequently, teamwork.
The cost of replacing a good employee is far higher than the cost of an AI tool.
Final thoughts
I am not advocating against AI. I really believe AI is a great tool that can help us and speed things up for us a lot. I use AI daily, and I feel productive when using it.
My main point here is that we should always be in control of it, not the other way around. We should criticize its outputs, learn the basics and foundations, and sometimes do things by ourselves, embracing the learn by doing process.
Do not delegate your entire career to AI, take ownership of your decisions. This path will require one of the hardest things in life: discipline. But don't give up on it, the more you exercise discipline, the easier it becomes.
Let's create more original content, thoughts, and ideas in an AI world where so much feels generic.
SwimAgainstTheTide
Top comments (0)