Understanding the problem is more important than the solution.

AI has completely devalued the solution and possibly you. The new wealth is in understanding. Go mine it.

Here’s the problem. In a pre-AI world, you could get by as an junior, mid-level and perhaps even senior engineer without understanding things beyond your codebase. There was enough value in your solutions and implementations to justify your salary. You didn’t need to add more value by being that uber-engineer who also had a super deep understanding of everything. You were paid handsomely for it. Until now.

That world is gone. The value you created by implementing solutions has evaporated. With a series of sufficiently targeted prompts and feedback loops, AI can spit out solutions all day long. But those solutions will only be of value if the person who created the prompt understood “everything” to at least an high intermediate level and preferably an advanced level.

Being able to derive a solution and an implementation is no longer the value. Developing (quickly) advanced understanding and being able to show it is the skill. The solution is now the substrate on which your understanding is laid.

That kind of sucks because for a lot of engineers the joy of the work was that everyday you’d learn as you worked. You were thinking and you were learning. That’s gone. You learn very little prompting AI day. But you do earn a lot of relationship capital if you and your work is visible and clearly adds something to the bottom line. Consider technical debt, here’s how clearing that debt used to go.

Example with a pre-AI solution and implementation focus:

“There’s loads of technical debt in the system, it’s killing team vibe and slowing us down.” probably, you, the engineer

“Ok, we’ll get to it after the next release, just keep going.” VP of something

Example driven be knowledge, understanding and relationships:

“There’s this complex part in the code where lots of On(2) operations are bottlenecks. I had the AI refactor it to On and deployed it out to a feature branch. All the integration tests are passed and it’s now in production. Response time by is improved by X% we’ve reduced the total cluster size by Y% and using smaller CPU sizes. Savings around £Z per year.” hopefully, you, the engineer

“Nice, that’s awesome and you’re awesome.” VP of something, who is being KPI’d to reduce operating costs and who also influences layoffs at the board level

You may not get to write the code and get free CPD everyday, but you do get to develop your understanding of an organisation and what drives the people in it. The better you can get at developing your understanding, the more desirable you are and an LLM is an excellent tool to help you do that.