AWS leader explores skills developers need in the AI era

At Arizona State University, partnerships with industry leaders help fuel innovation. A longtime partner of the university and premier sponsor of the Agentic AI and the Student Experience summit, Amazon Web Services (AWS) took the stage on Wednesday, Oct. 21, to reveal how AI is changing how software is built, who builds it and what it takes to succeed. 

Jeff Barr, AWS vice president and chief evangelist, described how AI is changing software development faster than many organizations can adapt. In this fast-paced environment, he said the success of next-generation developers is more about people skills than technical ones. He outlined how AI is reshaping developer roles, team workflows, data strategies and the relationship between technology builders and their clients.  

Barr urged developers to sharpen skills that have little to do with code: Observing customers closely to understand their needs, reading widely to learn how great authors communicate and writing clear, detailed prompts, since the better a description captures the desired outcome, the better AI can deliver it.

He emphasized that as AI speeds up coding, teamwork matters more than ever, with developers needing to communicate quickly and often, ideally working in the same place or within a time zone or two of one another. The most successful developers will be those who can become instant experts in whatever skill their team lacks. He also called for what he described as “M-shaped” developers, who have several areas of expertise rather than one deep specialty.
According to Barr, “The pace of change is such that developers need to learn something new every day: a new technique, a new model, a new version of a model, a new tool, a new way to use your model, a new way to interrogate your AI tools to learn more about your applications. A learning component must be part of your daily routine, your daily cycle.”

He also urged developers to stay comfortable with uncertainty and keep experimenting. “Ask it,” he said, is his shorthand for AI, because even model builders don’t know what a model can do until someone tries. He suggested retrying failed attempts months later, since tools keep improving.

Valued assets in software development are shifting as well, with code becoming disposable in some applications and specifications becoming the lasting asset. Since apps can be rebuilt quickly and need clean, well-described data to draw on, data is gaining importance for the same reason.
Barr warned that the greatest dangers surround the code, including testing, approvals, and deployment. Failing to keep pace with AI-accelerated development could leave organizations shipping faster but breaking more. 

Developers using AI tools can produce 10 to 20 times more code, but he said bug density remains constant, so older testing methods can cause production systems to break 10 to 20 times as often. His recommendations include testing earlier, using AI to generate test cases, validating specs, replicating production environments on developers’ laptops and tuning deployment pipelines.

He also noted that managing multiple AI agents can be mentally draining, with some developers reporting they are exhausted after four or five hours.

As Barr looked into the future, he reflected on his past, including his first programming job in 1982. “Back then, the best way to be a developer was to be very shy and quiet and to lock yourself in a room and never talk to anybody. The tables have now turned. Social skills are a path to success in today’s technologies.

For the second year in a row, AWS is the top sponsor for the summit. Additional programs include Valerie Singer, general manager of AWS Global Education, on the main stage, as well as a workshop and a breakout session.