There are no modalities that are not handled by AI. AI systems will reach even further, planning advertising and marketing campaigns, and automate social media posts. Most of this was unthinkable 10 years ago.
However, the first machine learning driven algorithm then performed the first step from the lab to the first product. They began curating content on YouTube and social media sites. They began recommending movies on Netflix and singing on Spotify. Ranked search results. They played strategic games on the same level as humans. General rise in AI-enabled thing It's spectacular.
Ai at work
And the workplace is not immune to this. As an undergraduate, I was studying how to build rules for hyperplanes, center of gravity, and backpropagation, but for most of my research, AI was primarily considered an orientation for academic research. This has changed a lot since I entered the job market. Employers and employees have recognized the possibilities of AI for their work. In most (digital) workplaces, AI is rapidly becoming an invisible colleague.
Many dedicated AI tools have already leapt into desktops. Programmers use AI-assisted coding tools, data analysts prepare pipelines from a single sample file via AI, and designers draft faster with AI-generated visuals. These tools undoubtedly make the work easier. But they also raise deeper questions:
What is your job?
What is my job really? Do I actually need to communicate something with my code in detail?
The more you apply the workflow, the less you need to engage in work materials. It may turn out we are not need To become an expert, he has deep knowledge of rather narrow topics, but rather shallow surfers, taking Ai-Glimpse everywhere.
In other words, we become mere managers of how AI does work. Before that, please note that there is no “us” work.
Is it possible that it is fulfilling? Does our work require a certain sense of depth?
I remember very well when I had to handle multiple simultaneous projects. Before AI became entrenched in the office, I often switched between three different projects per day and almost unrelated projects. With semi-urgent interruptions, you can imagine there is not much time to spend on a single topic. I had already had to switch over before I could go deeper into any topic to make real progress.
Today, AI systems often act as proxy, making sure you don't have to be involved in your project in the first place. We may only work on a single project, but we I will encourage future paths – It leads to questions:
If we work using AI, what is our job?
Is our job just doing more work? AI is often welcomed to help us do more. This means that given the same working hours, we need to be involved in even less material.
This means that by definition you cannot gain deep experience on one topic.
Furthermore, this means that, in principle, we can do work that is fully relevant to our skills.
Finally, it means that someone else can do our job.
Therefore, it can be swapped as soon as AI automation scales.
How can I prevent this?
We use AI intentionally. Think first and prompt later
In my opinion, the only way* is to use AI intentionally and selectively. Don't outsource your thoughts. Don't let the ability to think deeply and critically collapse without unconscious use.
Using AI tools for truly boring tasks that can be done by a neat and skilled person is completely fine. For programmers, using AI involves a safe (in the sense that it doesn't create a dumber) using AI involves summarizing the codebase, creating a README document, generating a boilerplate, or loading and cleaning data.
But if the task at hand requires trade-offs with human judgment, interpretation, or a particular design choice, then that's when you need to resist the temptation to hand it over. These are just the moments when you build the expertise that will keep you irreplaceable.
To make this more specific, you can use this simple heuristic when deciding to use AI assistance.
- Low stakes, repetitive, well defined tasks → AI can help.
Examples are formatting code, generating test stubs, and writing SQL queries. - Tasks that require high stakes, ambiguous or human judgment → do it yourself. Examples include designing a system architecture, interpreting experimental results, and making ethical decisions.
This rule of thumb keeps automating the “boring” thing while protecting the task of actually building expertise. To integrate heuristics into daily practice, you need to deliberately pause before the task. Ask yourself: Do you need to understand this in depth, or do you just need to get it done?
Next, if you understand the goal → start manually. Code the first draft, debug yourself, sketch the design. Thinking about it, you can extend your work with the output of an AI system.
However, if the target is simply an output → accelerate to AI. Prompt, adapt, and repeat with the next task.
Think of it as a mantra: “Think first and encourage later.”
Then, at the end of the work week, you can look back: which tasks did you outsource to AI this week? You did learn Do you want to complete something from those tasks, or simply complete them? Where did you benefit from being more involved?
Closed thoughts
As AI is increasingly used in the workplace, our real job may not be to generate more output with AI. Instead, our job is to directly engage the material when it matters – to build judgments, insights, and depths that the system cannot replace.
Therefore, use AI intentionally. Yes, it automates the boring parts, but protects the parts that grow. That balance keeps your work worthwhile, but also fulfilling.
* Non-equals for most machine learning people who spent a considerable amount of time building a career in data science: switch careers and do something manual and offline. Examples include construction work, hair dressing, waiting, etc.
