As AI assists coders, IT leaders seek more philosophical skills

Yeah, you know Java, but do you know…Descartes?

As AI tools speed up code development (or aim to), IT bosses and directors who spoke with IT Brew are increasingly searching for more abstract skills in their coders, like a way with philosophy and anticipating user behavior.

“Building things is easy, because of the technologies that we have today. Deciding on what to build is going to become more and more important,” Sharan Gurunathan, VP of engineering for the cloud solutions group at the IT services and consulting firm Presidio, told IT Brew.

The what requires some understanding of human need, or user need, according to Gurunathan.

The code ahead. Programmers have plenty of AI options for code assistance. Amazon’s CodeWhisperer or Microsoft’s Copilot, for example, offers suggestions based on comments and existing code. The Presidio team has been experimenting with how large language models like OpenAI GPT-4, Llama, Amazon Titan, and Claude can support software development needs, like code review, Gurunathan said.

With any free time afforded by AI, the Presidio VP wants to see his programmers think about the user.

“Someone who is sitting on a factory workflow might have a different span of attention, when compared to someone, like a production manager, who is sitting inside an air-conditioned room,” he said.

I code, therefore…Goldman Sachs’s Chief Information Officer Marco Argenti recently made the case in Harvard Business Review for why coders should study philosophy.

“One would have to first and foremost master reasoning, logic, and first-principles thinking to get the most out of AI—all foundational skills developed through philosophical training. The question, ‘Can you code?’ will become, ‘Can you get the best code out of your AI by asking the right question?’” Argenti wrote.

It doesn’t hurt to know coding, too. Steve Bennett, director of cloud-native applications at Soliant Consulting, looks for candidates with computer science and technical skills in React, Node, JavaScript, TypeScript, and SQL databases.

Bennett’s team uses AI-assisted tools occasionally like CodeWhisperer to speed up development processes and gain insights about code layouts.

“But certainly, we’re not at a state where we can just flat out trust AI-generated code without having to review it and to see what it generated,” Bennett said.

A study from Stanford University, released in December 2022, found that participants with access to an AI assistant based on OpenAI’s code-davinci-002 model “wrote significantly less secure code than those without access.”

Coding, experience. Mike Lempner, CTO at Mission Lane, helps to develop the fintech’s company’s systems, including mobile apps and payment-processing platforms. While an ideal candidate, Lempner said, has experience with financial services and multiple programming languages, the director also emphasized the importance of experience outside of coding and in less technical fields.

“It gives us an opportunity to make sure that we don’t just have a team of people who are coming at a problem with the same perspective,” Lempner told IT Brew.

Lempner’s financial experience, for example, often leads to questions of a coding project like: What are we trying to do from a business standpoint?

A coding quality that shines in any era, according to both Lempner and Gurunathan: people who are willing to learn, given tech’s constantly evolving nature. (Gurunathan, who sees willingness to learn as “first and foremost skill,” began his career as a mainframe programmer in Rexx, an early programming language.)

“What might have been popular five years ago, from a programming-language standpoint, may not be what we’re using today, and may not be what we’re using a couple years from now,” Lempner said.