Reality Check: Understanding the Truth about AI in Software Development

One of the great things about my job is I get to spend a lot of time talking to some of the brightest minds in the software industry – whether at conferences, in our training classes, or during our advisory and coaching opportunities.

By far, the most common question I get asked is “How will AI impact software development?” Frankly, the question is often asked with a sense of anxiety. If the current narrative is correct, the takeover of AI is inevitable and no job in any industry will be safe.

So how should our community react to this (not-so) new technology? According to software development expert Dr. Bob Martin, they can start by taking a deep breath. 

In a recent conversation, Bob and I sat down to discuss the reality and the hype around AI and its true impact. Long story short: The AI apocalypse is not near. AI will have an impact on how developers and testers work, but what the exact impact will be is still very up in the air.   

Here are some key takeaways from our conversation.

Artificial Intelligence, as it stands today, is not true intelligence.

Bob Martin says Artificial Intelligence is somewhat misleading. He instead calls today’s large language models and other “AI” technologies, ‘resourceful’. While AI technologies can pass the Turing Test (Alan Turing’s assessment of a machine’s ability to exhibit intelligent behavior at the same level as a human) in short windows, they currently cannot pass the Turing Test on a consistent or long-term basis. For this and many other reasons, we deem AI as not humanly intelligent.

AI will NOT take developer jobs.

“I can’t imagine a job function that is in jeopardy.” Bob Martin says he doesn’t believe AI’s emergence poses a threat to developers or other software experts. Bob says he doesn’t see programmers getting replaced by AI or the number of programmers being diminished. Instead, he thinks programmers will find ways to make themselves more productive as a result of AI. 

AI still needs help in supporting modernization.

While AI does have the potential to support organizations who are modernizing their code from legacy systems like COBOL, Bob Martin says he doesn’t see today’s current AI technology as being able to fully understand different languages and meaningfully apply updates. I also mentioned that today’s LLMs are largely informed by open source coding languages and systems making it difficult for AI to understand closed source code like COBOL. While organizations may consider AI to help translate between languages, it’s essential for developers, testers, and other experts to thoroughly examine AI’s output to make sure modernization efforts are carried out effectively. 

What do you think? Do you agree or disagree with our thoughts about the emergence and impact of AI? Whatever you think, it’s critical for us all to understand this rapidly changing technology as best and as fast we can. Watch the full conversation here.

Further AI Training

I’d suggest starting with our latest Coveros certification training course: Fundamentals of AI. This introductory course introduces you to the what, the why, and the how around AI and software development. After completion, you will earn ICAgile’s brand new Foundations of AI certification.