Being a fresher in the age of AI

Young engineers who are entering the workforce now or have entered in the last 2-3 years, if you try to look behind all the token maxxing high, are facing a grim reality right now.

I use the word “high” very much intentionally because this is how the AI subscription systems have been designed. If you are at the beginning of your career and funding your own subscriptions, then to even stay competitive, you have to keep paying every month. And it requires active flow of income just to keep your subscriptions alive as the free tiers are mostly not great.

Everyone around you is telling that if you don’t use AI then you are not being useful, or not doing a favor to yourself, or that you are falling behind everyone. So you are under this huge stress to just keep using AI more and more to churn out more code/outcome to show that you can be useful when you join a job. The focus is not even close to learning the first principles of programming.

I have been telling everyone that this AI buzz will hurt the young people the most because they will bypass all the pains of incremental progress. Experience in programming has always been about how many errors you get to debug yourself. We have delegated that whole process to AI. Folks who have been programming before the AI era mostly know a good or bad code/command because we suffered the bugs. If an AI tool spews out bad code/command, we can immediately smell it. But for young engineers, that’s the smell of code they know. They have to go with whatever is being presented to them. And keep doing the same over and over again.

So the more tokens they use, the more they have to use. With no exit hatch. Sounds very similar to some other addictive habits we know, right? Hence the use of the word “high”.

Now at the flip side of the job market, almost every single job post I see has 3000 applicants because there are too many people, and too few jobs. And also, because almost nothing is going through to the recruiters because they don’t know who to trust and how to filter the 3000 applicants. They don’t know how to figure out which resumes are written by AI and which are written by humans. Who actually knows their stuff, and who doesn’t. And if they even make a call of picking 100 among them, they don’t know who will cheat in the interview using AI, and who won’t.

From the applicant’s side it’s a brute force process. From the recruiter’s side it’s a “guilty until proven otherwise” mentality because of all the bad apples. No space for trust left in the system.

If you are a young engineer who is facing this on day to day basis, my advice to you would be to focus on first principles. Look behind all the token maxxing. If you see your peers generating AI slop day in and day out, then calm yourself and see what they are learning from it instead of what they are generating. Because that’s what is gonna matter in the long run. This AI buzz is gonna end and those who know how computers work, how databases work, how servers work, how internet works, how containers work, how distributed systems work, only those will be able to stick around.

You can get into a job by using AI. But when the AI slop breaks and you are drowning in the code that does not make sense, first principles will save you. When an interviewer asks you a question your agent didn’t ask you to prepare for, your first-principles knowledge will save you. Go learn the basics.

If you are reading this and have questions, please reach out. Don’t lose hope.