𝗪𝗵𝗮𝘁 𝗜𝗳 𝘁𝗵𝗲 𝗖𝗼𝗱𝗲 𝗗𝗼𝗲𝘀𝗻’𝘁 𝗠𝗮𝘁𝘁𝗲𝗿 𝗔𝗻𝘆𝗺𝗼𝗿𝗲?

Most programmers aren’t very good at writing requirements.

It’s not because they’re lazy or careless. It’s because 𝘄𝗲 𝗱𝗼𝗻’𝘁 𝘁𝗲𝗮𝗰𝗵 𝗶𝘁.

In most CS/EE programs, the focus is on algorithms, data structures, and writing code that works.

Nobody teaches you to say: “Let’s discuss how to describe what the system 𝘀𝗵𝗼𝘂𝗹𝗱 do clearly, unambiguously, and completely.”

Programmers often don’t develop that skill until they’re senior engineers.

But here’s the twist.

We may be heading toward a future where your ability to code matters less than your ability to specify what the code should do.

𝗙𝗿𝗼𝗺 𝗔𝘀𝘀𝗲𝗺𝗯𝗹𝘆 𝘁𝗼 𝗔𝗜: 𝗔𝗿𝗲 𝗪𝗲 𝗥𝗲𝗽𝗲𝗮𝘁𝗶𝗻𝗴 𝗛𝗶𝘀𝘁𝗼𝗿𝘆? When higher-level language compilers first appeared, many programmers were skeptical.

They pored over the generated assembly, shaking their heads:

“𝘐𝘵’𝘴 𝘯𝘰𝘵 𝘰𝘱𝘵𝘪𝘮𝘢𝘭. 𝘐𝘵’𝘴 𝘪𝘯𝘦𝘧𝘧𝘪𝘤𝘪𝘦𝘯𝘵. 𝘐 𝘤𝘰𝘶𝘭𝘥 𝘥𝘰 𝘣𝘦𝘵𝘵𝘦𝘳 𝘣𝘺 𝘩𝘢𝘯𝘥.”

And they were often right…at first.

As compilers matured and hardware got faster and cheaper, priorities shifted. They stopped obsessing over machine code and focused on the what and why—not how every instruction worked. Most stopped looking at the assembly output altogether.

Fast-forward to today.

We’re hearing a similar refrain about AI-generated code. When developers use AI to “vibe code” (expressing intent in natural language), the critics are quick to respond:

“𝘛𝘩𝘪𝘴 𝘤𝘰𝘥𝘦 𝘪𝘴 𝘴𝘭𝘰𝘱𝘱𝘺, 𝘐𝘵’𝘴 𝘯𝘰𝘵 𝘦𝘧𝘧𝘪𝘤𝘪𝘦𝘯𝘵, 𝘕𝘰 𝘳𝘦𝘢𝘭 𝘥𝘦𝘷𝘦𝘭𝘰𝘱𝘦𝘳 𝘸𝘰𝘶𝘭𝘥 𝘸𝘳𝘪𝘵𝘦 𝘵𝘩𝘪𝘴.”

It’s the same tune, different instruments.

If AI keeps improving and hardware keeps getting faster and cheaper, we may be heading to the same place:

One where 𝗺𝗲𝗲𝘁𝗶𝗻𝗴 𝘁𝗵𝗲 𝗿𝗲𝗾𝘂𝗶𝗿𝗲𝗺𝗲𝗻𝘁𝘀 matters more than how clever the code looks. Maybe programmers will even stop looking at the code all together.

𝗦𝗼 𝗪𝗵𝗮𝘁 𝗗𝗼𝗲𝘀 𝗧𝗵𝗮𝘁 𝗠𝗲𝗮𝗻 𝗳𝗼𝗿 𝗬𝗼𝘂? If you’re not great at writing requirements, you should work on that skill 𝗔𝗦𝗔𝗣. Take a class, read a book, offer to help with the spec.

If the future of software really becomes telling an AI 𝘄𝗵𝗮𝘁 to build, then development 𝗥𝗘𝗤𝗨𝗜𝗥𝗘𝗠𝗘𝗡𝗧𝗦 𝗔𝗥𝗘 𝗧𝗛𝗘 𝗡𝗘𝗪 𝗖𝗢𝗗𝗘.

Even if that future never fully materializes, learning to write clear, testable, complete requirements will still make you a better programmer:

• You’ll think more clearly.
• You’ll design more robust systems.
• You’ll collaborate more effectively.

Whether leading a team or prompting an AI, you’ll benefit from speaking the language that matters most: 𝗧𝗵𝗲 𝗹𝗮𝗻𝗴𝘂𝗮𝗴𝗲 𝗼𝗳 𝗶𝗻𝘁𝗲𝗻𝘁.

• By @Greg Burk, cross-posted from LinkedIn

#ai #requirements #WolffElectronicDesign