Software Engineer jobs (Mid-Senior) can NEVER be replaced by a Coding 'AI' like Devin

The primary function of a Software Engineer is not just writing the code. Our primary function CANNOT be replaced by "AI" code generators (if it could, we wouldn't have needed all the Agile/ProjectManagement overhead for all these decades).

I said what I said.

The video I mentioned at 5:49 about how software is getting worse (and explaining the graph towards the end of this video) is here:
Software is bad, and it's been getting worse. But it's still worth doing.

NOTE: (Not that anyone will read this, but) When this video uses the term "AI." it (like most of the current discourse these days) means current and near future "Generative AI" a.k.a. Large Language Models. If you're thinking about the Agents from the Matrix or the Robots from Terminator, that's DIFFERENT. That's the OTHER AI, the Human-Intelligence General AI, Sometimes called "AGI." That DOES NOT EXIST. For the time being it's PURE FICTION. It may get here soon, although I (and a lot of people smarter than me) seriously doubt it, but we'll see.

[Baditow]

as a software engineer, I was ready to challenge your arguments, but after listening it seems you've covered all the bases. well said and good video. one day within' a shorter period than we probably expect, the ai is likely going to become autonomous and exponentially build in a way that no human or amount of humans would ever be able to keep up; though, at that point, we're not talking about software engineers but the concept of employment as a whole.

[StoCoBoLoMC]

"The job of the Software Engineer is more about people, than it is about code" Perfectly said!

[MrTheEPICman]

I never, ever comment on YouTube videos. Take this as my utmost thanks for having a balanced and sane opinion. Everywhere I look, people seem to be either certain AI will kill the software industry very soon, or that it will be nothing to worry about whatsoever. Hardly do you find someone who doesn't think in absolutes, but breaks down the situation at hand to the best of their knowledge and comes to a sound conclusion rather than jumping to one. I will be graduating with a bachelor of honours in software engineering by the years end, and it's an absolute breath of fresh air to hear a logical argument on the future of our discipline rather than fear mongering or blatant disregard. I am looking forward to your future videos.

[jamessullenriot]

Ahhh, so refreshing to see someone that actually knows that developers are more than leetcode pros trying to get into FAANG.

[RainOnline]

I love this rumor. Less people will be prone to starting a software engineering career. Therefore I will be more valuable.

[mangopeepl]

Any juniors breaking into the market right now, listen up!

Start by building a suite of self made products all by yourselves or with a group of friends. Take it online. This way, you have a portfolio that show that your're resourceful enough to do much more than what a junior in 2010-2020 could do. A full blown MVP or two shows a recruiter that you're familiar with the basics of software development, DevOps, Testing etc. This implies that you are ready for the people skills and experience required for senior roles.

As it is, juniors in most companies simply write repurposed code typically copied from stack overflow anyway. This level will be eliminated(as chatpt/copilot already does this) and juniors will be given the tasks that mediors are given these days. Mediors will be given senior tasks and seniors will be forced to take on more business roles to survive.

My suggestion to all software developers. Take up some business/design courses. Learn people skills. You can no longer be the loner nerd that sticks to coding.
YOU WILL NOT SURVIVE. Unless Of course, you happen to be in the top 10% and are prepared for the immense competition coming your way.

[Bregylais]

As an oversimplified mathematical model, I like to imagine the relationship between all of these roles as an multiplicative equation, where ever role involved (each multiplicand) contributes a value in the range of 0 to 1. The goal is to build a product (pun intended) of the value 1. During a perfect sprint, when everyone brings their A-game of 1, this goal is realized. But this is not the world we live in. Product managers miss-communicate with the stakeholders, developers take shortcuts with the implementation, the testing-team misses a crucial edge-case. As soon as even one involved person only brings a 0.9 to the table, the feature itself suffers and will deviate from what some dreamy stakeholder envisioned in the beginning; and it gets only worse with more than one person messing up on any small or big work detail.
The chance that we'll ever get to a point where a stakeholder can talk directly to a software development LLM (which in turn has to take all these mentioned positions) and gets *exactly* what they were hoping for seems slim at best, certainly not within the next 10 years. I wholeheartedly agree with your points made in this video. New sub, btw!

[eugenefritz]

That's why following trends often leads you to rubbish.
Thank you Carl for your insights and transparency.
You're sublime lad.

[SisarothSC]

I feel like I live on another planet when I watch a video about AI coding and everyone is doomposting about how they are gonna lose their job soon. You would think people are all in on nvidia stock and are just shilling.

Even not considering the human aspect of being a developer I still have my doubts, sure AI will be an amazing tool to speed up development. But doing it all seems hard to believe for anything but the most basic applications.

5 years ago Microsoft was gushing over PowerApps and how it would allow non-technical people to create applications. I did develop some PowerApps but as soon as you have to implement a functionality with any kind of complexity, it quickly becomes as complicated as normal programming.

[johnsmith-rotw]

Look at it this way : modern planes such as A350 or A380 are largely automated. In fact, they can auto lend. Does it mean we have self piloting planes yet ? Nope. Plane pilots are still as important as they were 20 years ago. Plane pilots still go through the intense training they have always been through. Many routine things can be delegated to the auto pilot during the flight, but what if something goes wrong ? A system failure or a big storm ? Something so bad that the auto pilot becomes utterly useless ? The same thing can be said for surgeons who can today rely on sophisticated machines to help them. Those machines are not replacing surgeons any time soon either.
AI in SWE should be seen the same way. It can be a convenient tool for an experienced engineer who understands what he's doing. It helps him boost his productivity. The SWE does not blindly ask the AI to do his job, he only gets it to do things that he knows about, but would take him longer to do himself. He monitors what the AI does, and is ready to correct course when it inevitably goes a wrong way. Whoever thinks of AI as a way to reduce the size of the team is in for a rude awakening, IMHO. This said, with the high turnover we see in typical tech companies, the PM who got the genius idea to replace a SWE with Claude 3 or whatever, will have been rewarded for his genius idea, then will have changed job before the company pays the price for that "genius" idea.

[frobro]

*Really really good video* . I have been really concerned about this but you put it perfectly and it *matches my experience* that mid to senior level software really is more about people than code. That is exactly who I worked with or bumped heads with on every software project I ever did. At the beginning, middle and end there was a ( _sometimes incredibly_ ) non-technical person who had to understand and sign off on the design, writing and verification of everything so the work would be accepted not a machine.
Edit: I thought about it and I recalled a book that I think backs this video up well called "Code Simplicity" by Max Kanat-Alexander.
Especially, Chapter 2 "The Purpose of Software"; hint about the subject for the chapter, it's people.

[bouihabdelouahed]

Good counter-arguments. Pretty balanced.

Thanks, man.

[bwhit]

What if project managers get smarter? Suppose one person was subject expert and product manager and understood the business logic. They could imagine the rough architecture of the software and then use AI to turn their rough sketch into reality. They understand the entire company/software/product in a vague sense, but they leave all the technical details to the AI.

Assuming AI gets a whole lot better, of course. You make a good point that we can’t just substitute AI into current companies, but what if we design new company structures to maximize the potential of AI?

[jackblurry]

your face at 5:47 made me laugh because you looked like a person that had brief flashback of 20 years of fixing horrible code lol

[red_gandalf_chegando_atrasado]

I am glad I got to this channel early 😎

[jlecount]

That's really solid and insightful. Enjoyed the video - thank you!

[drummermike]

Wow, love your content! Changing my mind with every video on how I feel what AI is going to do to our industry.

[frankie_goestohollywood]

I hope you’re right - your breakdown makes a lot of sense 👍💪🤓

[yashpreet]

Thank you for sharing your thoughts and knowledge.

I really appreciate it.

Keep moving forward

Best wishes, have a good day

[KuzzatAltay]

I've been the CEO and co-founder of a coding bootcamp for nine years and have been in this industry for 12.

I just want to say that your videos are incredibly valuable during these challenging and confusing times for many people. While no one can predict the future with 100% accuracy, your knowledge and insight are particularly useful, especially when the majority of discussions about AI do not come from those with backgrounds in AI or computer science.

I watched three of your videos this morning and plan to watch the rest. A big thank you. With the authenticity of your content, this channel deserves 100, 000 subscribers.