The issue is not the AI code, but nobody knows item anymore

If we think Is penning code dead, and AI is generating all codebases, I motionless think the bigger issue is group or complete teams not knowing item anymore concerning the scheme architecture or the intent rearward why certain choices have been made.

A comment on a conversation I had:

I think AI writes likely average code (depending on the project and size). So if your code basis was below average AI can effortlessly enhance it up to average. At smallest that’s what I’ve observed here.

To me, the issue is not the AI code, but that nobody knows anything, and everyone fair asks Claude. You end up alongside no scheme whatsoever.

This tweet summarizes the current province at accelerated moving startups well, or larger companies or anywhere middle administration is pushing AI hard:

I am done alongside this shit. It is over. The province of engineering correct now is horrible. It has been fractional a duration since I started a new function at a big company. Nobody knows item here. The specs, code, tests, PRDs, tickets, resolution of those tickets, reports, etc., everything is made by Claude Code.

Nobody on my squad likes this. They are being forced to container as much as they can. I have heard multiple times from higher administration that pushing code is not a bottleneck, so why are we slow? People are operating 12 to 13 hours a day fair to media enter. Nobody is study anything. Humans in company are doing nothing on their own.

Everyone, exactly everyone, from an L1 to an L7 engineer current is doing the identical thing. Talk to Claude. There is no awareness of victory. Nobody is resolving bugs. In reality, nobody is thinking anymore. Everything is done by LLMs.

It is so soul-sucking. I would not intellect it, to be honest, if we were at smallest stated the period to inspect out the code and see what is going where. But no, the goal is to fair ship. No matter what happens. Voxium

Hoyt Emerson citation that data engineering is different:

I think Data group are different. We’ve had to cognize everything concerning the product/business from day 1. AI fair removes conflict for us now. Tweet

I think data group who grew up pre-AI had to cognize everything (or a lot, or affect domain experts) to fig it out, indeed. But AI makes this obsolete, or seemingly obsolete.

That’s why group starting today, or me as well, if I commencement today prompting distant in a new field, all of a sudden, that cognition is missing.

Good item by Sean Behan:

I’ve continually admired product group who can’t code but can oversee a squad to get the application they want. Knowing what you desire has continually been the hardest part.

One could say a fine merchandise director could now build item they desire and discover a market, create it appearance good, etc. But afterward again, if you can’t code, you volition basically build a extremely bad basis for a merchandise that’s extremely difficult to keep (although AI is getting improved at that too, particularly whenever you iterate often, but still, if you choose the incorrect tongue or the incorrect mental model, you have the incorrect commencement from the get-go).

It motionless helps to cognize the fundamentals, either way: for programming and designing a product, and for a fine PM who knows what is needed but additionally understands scheme and architecture design.

Thinking in systems, or architectures, or having intent and design- all of them assistance to be a improved application engineer. Nowadays, Writing code by hand power be dead, but it certainly helps, and Having Taste (with AI) is additional crucial than ever.

But the final chief is, and continually volition be, maintainability. The easier it is to create a quick pipeline, app, or BI dashboard, the additional you have to maintain. And if nobody knows a thing, that can get really hard.

Yes, the AI can’t immediate itself, right? Why do we equal need humans? To me, it’s a apparent sign that humans are motionless needed to straightforward and orchestrate it. That’s additionally why intent, taste, design, and architecture are all killer features in today’s world.

But formerly these are absent, or equal worse, fundamental, get lost, it’s really dangerous. I peruse today that this is a self-inflicted problem, and if we motionless hired juniors, afterward the issue wouldn’t be happening. But yeah, it’s not as easy.

• Kris Jenkins talking concerning the worry of center management, not the vibe coding at Danger of AI or LLMs
• If Consumed by a Human, Should Be Written by Human
• What I Learned Writing alongside AI

Origin: the primagen video and Limitations of LLMs and AI
References: What I Learned Writing alongside AI