Promises and Perils of AI

I had the opportunity to record a podcast with the CTO for the software integration and digital consulting practice I work in. When I joined this side of the business, I felt both gratitude and intimidation for the opportunity to walk among giants, especially the CTO. He has developed software for as long as I’ve had a computer (he started with an Apple II where I started with a Commodore VIC20) and has made his career accelerating the adoption of cutting-edge patterns and practices in software development.

As CTO, you act as both trendsetter and trendspotter. I asked about what trends he wants to see our practice drive and customers to adopt. Additionally: what trends he sees coming that we should prepare for. Without hesitation (or surprise), the conversation pivoted to AI.

We see organizations large and small already integrating AI into their business in a variety of innovative ways. We’ve built an entire service offering to help our customers integrate AI with their call center applications. Rather than the “Say ‘new’ to make a new reservation…Say ‘existing’ to change an existing reservation…” of IVR, a caller carries on a conversation with an AI-driven virtual agent which makes a more intelligent call routing decision while also analyzing the sentiment of the caller.

We already see organizations increase productivity while also enhancing customer engagement with AI. Software engineering represents another area where AI has the potential to disrupt. Generative, LLM-based AI can generate functional code. I kid my friend with a career in software engineering that AI will replace people like him. While I don’t actually see that happening, since the discussion with my CTO I believe AI will disrupt the software engineering profession in another way.

AI will not directly replace software engineers but rather make them more productive. While ChatGPT and other tools can write fully functional code, it still requires a software engineer to prompt ChatGPT in just the right way to make it produce something useful. A software engineer also needs to snap that piece of code into the structure of their overall solution architecture. As I went through the trial and error necessary to launch this site, I leveraged ChatGPT to produce some sample configuration files that I then customized for my own purposes. That saved me considerable time.

Therein lies the peril of AI. Some (but not all) software engineers will embrace the use of AI to help them write code and increase their personal productivity. Others will resist using AI and prefer to write code themselves. Those that do not embrace AI and the boost of productivity it promises risk seeing themselves left behind those that do.

“If you don’t like change, you’ll like irrelevance even less.” —Eric Shinseki