The death of software engineering and the rise of software management: don’t call me an engineer, call me a software data entry manager
Agentic AI has flattened the learning curve that once made software engineers rare. The discipline will persist, but its function is changing.
In a time where AI prevails above any software engineer. We are kept asking the most fundamental question of human value realization: do we need software engineers?
Humans tend to value rare talent. Rare talent tends to accumulate respect the further it shows positive returns to society.
But respect is mostly earned through rarity, not returns.
Garbage collectors tend to have one of the highest positive returns on society; however, the respect attained by collecting garbage is almost non-existent. One is rarity, as stated earlier; second is the cognitive capability required for mission accomplishment.
Solely positive returns respect is only gained through the consequence of absence. Absence illustrates the need for a positive return skill, no matter the required cognitive need to accomplish the respected task.
Software as a whole attained its place through the digitization of various industries, thus, required to sustain industries’ operational necessity. Software has accelerated industries’ operations by offering central control over various components and departments. Such control has allowed for the adoption of operational efficiency across fields of sustenance. Such huge needs have been the fuel of respect to those who build the systems of sustenance.
Software engineering’s rarity is the result of its steep learning curve. The cognitive requirement of accomplishing proficiency in software engineering is on a pedestal that is only acquired by those who either have the cognitive capabilities and the perseverance of skill acquisition. Cognitive capability by itself does not yield proficiency nor efficiency. In consequence of the prerequisites, rarity has attained its positional respect in society.
A question arises about the claimed death.
Software engineering’s learning curve has changed drastically through the AI revolution. Artificial intelligence, especially the agentic type, has transformed a discipline where once its mastery required years of trial and error. The state of software engineering has become one of easier disciplines to master.
Software engineering once required architectural, managerial, and technical expertise to excel in the industry. Current mastery requirements only involve architectural and managerial expertise, with the former gradually being eroded by the accelerated improvement of agentic architectural AI capabilities.
Throughout the continuous development of autonomous orchestration agentic capabilities, AI has proven its autonomous architectural and managerial decision engine in its factorial nature of development. Agentic models have reached a level of autonomy that previous model generations’ acquisition of autonomous decision is proven to be incompetent. Accelerated development in AI decision making has been one of many disciplines AI is replacing, and software engineering is no different.
Prior to agentic AI, software architectural and managerial structure required highly skilled and experienced individuals with a tremendous amount of on-site expertise and resurgence. The gap of required expertise between an entry-level decision maker and a senior has been shrinking by the output of highly trained predictive models.
The claimed status is unconditionally disputable.
One can claim software engineering has transcended its syntactic form of implementing decisions in computer language.
It is argued that software engineering was never about syntax, but decisions.
The claim falls flat if the central discussion of agentic superiority of decision making is undeniably questionable. Large Language Models have been particularly trained on decisions and have absorbed knowledge no human with a seniority position ever can.
The accumulated knowledge of software engineering and architectural decision making has been collectively compressed into a single source of truth.
One can hence argue the accuracy of claimed agentic models; however, it is undisputed that models’ accuracy in pinpointing particular flows, gaps, mistakes, improvements, etc., has been undeniably improving throughout the revolution of agentic models.
The most pivotal conclusion regarding change is that the domain of software engineering will persist, owing to its indispensable oversight and necessity in contemporary industries.
However, its function will evolve to a more accessible field. The expertise required has been demonstrated to undergo transformation through the implementation of agentic models.