Coding and System Design Skills for the AI Era
As AI advances, the key advantage will not come from asking it to do everything at once, but from designing systems that use its strengths while assigning other work to code or people.
As the AI era arrives, we need to consider how to interact with AI. Many people try to use today’s AI as though it were the AI of one or two years from now, and they often fail to get the results they expect.
Simply asking AI, “Please do this for me,” is not enough. To obtain the desired result, work must be broken down into small tasks that AI can handle. Parts that can be solved with simple programming should be handled separately, and human intervention is sometimes necessary. In other words, using AI requires systematic design skills.
What system design means
System design is the ability to divide a task into work that AI can handle, work that simple code can solve, and work that requires human involvement, then arrange those components effectively. It is especially important to minimize human intervention and expand as much as possible the area that can be handled by simple code. Coding experience and project experience are essential for doing this.
Someone with these capabilities can use AI to automate literature searches and bibliography management, or build a system that automatically produces review papers. Someone without them may only hand everything over to AI and be disappointed by the outcome.
A widening difference in AI use
In one or two years, everyone may be able to use more capable AI more easily than they can today. However, people with system design skills will be able to achieve even more remarkable results. Until AI development reaches a point of stagnation, differences in the ability to use AI will continue.
The gap between a person who says, “AI, do this whole thing for me,” and a person who designs a system that maximizes AI’s advantages is unlikely to narrow easily.
The jagged frontier of AI capabilities
The concept of the “Jagged Frontier of AI Capabilities” is useful here. AI performs exceptionally well in some areas while still showing clear limitations in others. We need to identify what AI does well and focus on those areas. This will be a core strategy for navigating the AI era.
Figure source: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4573321
AI development is accelerating, but merely keeping pace with it is not enough. The real winners will be those who understand AI’s capabilities accurately and can design systems that use them effectively. At this point, how we grow alongside AI matters more than ever.