Will AI Dominate Programming? Insights from Microsoft, IBM, and Salesforce

The overall situation is that AI’s rapid development is driving a major transformation in computer programming, with industry leaders holding divergent views on how AI will reshape software development; some see a revolutionary shift while others are more cautious. A Microsoft executive is optimistic, predicting that AI will dominate the programming field within the next five years.

Microsoft’s Chief Technology Officer Kevin Scott predicts that by 2030, 95% of programming code will be generated by AI, but he quickly clarifies that this does not signal the end of human involvement in software engineering.

"This does not mean AI is doing software engineering work… the author will still be human," Scott explains. "It creates a new abstraction layer, moving us from being masters of input (programming languages) to masters of prompts (AI coordinators)."

He does not believe AI will replace developers, but it will fundamentally change workflows.

Software engineers will increasingly rely on AI tools to generate code from prompts and instructions rather than painstakingly writing each line. In this new model, developers will focus on guiding AI systems instead of manual coding, using prompts to express their needs while AI handles repetitive tasks, allowing engineers to concentrate on design, problem‑solving, and higher‑level responsibilities.

Despite the bright outlook, current AI systems have significant limitations, especially in memory. Scott admits that today’s AI assistants are “too transactional,” struggling to recall past interactions and adapt to user preferences over time.

He believes this is a temporary issue and predicts future AI tools will learn from previous interactions to provide more personalized and context‑aware assistance.

"Where agents have memory, the capacity is limited," the Microsoft CTO says. "Next year AI’s memory capability will improve dramatically."

The industry is debating AI’s effectiveness in programming.

Recently, IBM CEO Arvind Krishna cautioned at the SXSW conference that AI will boost productivity rather than directly replace programmers. He estimates AI will handle about 20%‑30% of coding tasks, emphasizing its limitations in tackling more complex challenges.

"If you can produce 30% more code with the same workforce, will you write more or less?" Krishna asks, arguing that efficiency gains will spur innovation and market growth rather than cause unemployment.

Salesforce CEO Marc Benioff agrees with Krishna’s view, stressing the need to retrain staff for an AI‑driven future. He notes that because AI tools increase productivity, his company might stop hiring traditional software engineers by 2025, but he underscores the essential role of human expertise in collaborating with these technologies.

Meanwhile, Anthropic CEO Dario Amodei offers a more aggressive timeline, believing AI could generate up to 90% of code within six months, highlighting the rapid progress of these tools.