AI Won't Fix Broken Software. Better Thinking Might | Opinion

The most significant error we can make regarding AI is mistaking additional software for enhanced software, or presuming that what currently exists can be improved by merely incorporating AI. I've observed organizations hastening to affix an AI label to products that have undergone minimal changes, as though the appropriate acronym can metamorphose a routine feature into a revolution.

Some of this technology is exceptional. Some of it is merely outdated software adorned with a novel tag, and the disparity will have significant ramifications for the enterprises, employees, and communities navigating this transition. The systems, at best, are cumbersome, and at worst, are outright hazardous. The fervor surrounding AI supplanting everyone is starting to wane, but the fundamental transformation has merely commenced.

Surprisingly, nearly nine out of ten respondents to McKinsey's 2026 global survey reported that their organizations regularly utilized AI in at least one business function, whereas 44 percent stated AI was being expanded across the entire enterprise. The market is progressing toward a more challenging stage of the cycle, where enthusiasm must withstand scrutiny from budgets and measurable outcomes.

This marks precisely where the discourse ought to unfold. AI should augment human capabilities. Superior technology amplifies humans; everything else is superfluous. Envision what this could signify beyond the technology sector. A modest enterprise could acquire capabilities that once necessitated an entire department. An advisor could serve thousands of clients while preserving the discernment and rapport only a human can offer.

AI-assisted drug discovery is already yielding tangible outcomes. A recent milestone was a randomized phase 2a trial of rentosertib, a therapy whose biological target and molecule were identified via AI. The potential is immense, which renders the current wave of AI marketing more than an annoyance. It threatens to obscure what truly matters.

Software is entering an era where a collection of disjointed features will increasingly prove inadequate. I anticipate the most robust products to deliver a comprehensive service out of the box, possess the necessary infrastructure to ensure it functions correctly, and integrate seamlessly with the systems surrounding it. AI should execute useful tasks within that system, discreetly eliminating the repetitive tasks that waste human time.

I witness this already in sales organizations. AI agents can transcribe conversations, archive records, highlight pertinent context, and keep a team progressing without compelling salespeople to spend hours maintaining a CRM. The humans retain responsibility for relationships, judgment, and decisions. The machine manages the administrative burden that people have endured for years due to the absence of a better alternative.

This is also why I anticipate AI to become increasingly specialized. The notion of a single colossal corporate brain controlling everything appears impressive, but businesses possess disparate data and distinct requirements, and the requisite infrastructure simply does not exist and will not for some time. Purpose-built models will have specific duties, communicate with other systems, and collaborate alongside humans.

Commercial viability will ultimately dictate which technologies endure once the novelty fades. Conversely, there exists a darker potential for this future. Companies may continue to stack software atop software, affix AI to flawed processes, and leave employees maneuvering through an increasingly convoluted maze of systems. Expenses escalate, data becomes less reliable, and the technology intended to save time generates more work.

The promise of AI then transforms into its own source of weariness. AI can fundamentally alter the economics of work by lifting the ceiling on what a small team can achieve, compressing the cost of building and operating a business, and providing individuals with more capacity to apply judgment, creativity, and expertise. The value can be discovered where human capability can finally scale without every increment in ambition necessitating a commensurate increase in cost or complexity.

This alters the discussion about employment. AI will diminish the labor required for specific tasks, and businesses may be able to expand without adding personnel at the same pace. We must be candid about this. We must also acknowledge what transpires when companies resist adaptation. Businesses that cannot compete will eventually dwindle or vanish, taking jobs and economic activity with them.

Simultaneously, the barrier to creating something novel continues to diminish. Cloud infrastructure has already revolutionized the economics of launching a software company. AI is propelling that curve further. The next generation of entrepreneurs will be able to construct with minuscule teams, test ideas affordably, and vie for markets that once necessitated immense capital.

As some jobs become obsolete, new ones may emerge, but the overarching impact on employment remains uncertain. The genuine challenge for business leaders today is more immediate than forecasting some distant AI future. Examine the business you operate presently and identify the process people detest, the work that consumes hours without generating meaningful value, or the information that vanishes into a system nobody wishes to maintain.

Begin there. Automate it effectively, measure what changes, and employ what you learn to confront the next challenge. In essence, it must be impactful today while positioning businesses and consumers for the world of tomorrow. Software companies bear an equally clear responsibility. Construct products that resolve complete problems.

Make them useful from the outset. Offer people time and capacity back. Ensure the technology earns its place within the organization through the value it generates. The companies that proficiently make these choices could gain a substantial advantage because they will operate with less friction and provide talented individuals with more room to perform work exclusively humans can accomplish.

The future of software is already taking shape within those decisions, and the businesses making them now will wield a powerful influence on what lies ahead.