In this episode of "This is AI," hosts Erik Brown, EJ Johnson, and Ryan Elmore, along with special guest Alex Cruikshank, explore the transformative role of AI in coding. The team delves into how tools like GitHub Copilot are not just aiding but reshaping the software development process. They examine the delicate balance between leveraging AI for efficiency and ensuring foundational coding skills are not overshadowed by automated solutions. From enhancing developer productivity to redefining the coding landscape, this episode provides a deep dive into the dynamic world of AI and software development.
There's going to be efficiencies gained by these tools. AI as a co-pilot to assist engineers is going to produce efficiency. And it looks like it could be 30%, maybe in the best case. So that's a big deal. I also think there are classes of development that are going to go away because of AI. It's like there's things that we do now that we're just not going to do in the future.
In a recent boardroom discussion at a PE firm, surrounded by 10 to 15 executives, a pointed question was raised: 'Why should we continue to invest in software companies when tools like ChatGPT can seemingly replicate their functions?' This question reflects a common misconception. While AI like ChatGPT is powerful, it cannot replicate the intricate layers of intellectual property, the decades of accumulated data, or the millions of lines of meticulously developed code.
It's going to make developers more efficient. That’s going to be different depending on the seniority of the developer, the type of work they're doing, but hopefully it allows them to focus on the stuff that's actually germane to the business logic that they're writing, right? Rather than the scaffolding, the tedious development activities that we have to do right now that we tend to make mistakes on because it's not that fun, it's not that exciting.
So, Alex, you and I have talked previously about exactly what Brownie is talking about, where you put something like ChatGPT or custom OpenAI integration or GitHub Copilot in the hands of a senior developer. You could, depending on the task, depending on the work to be done, you could see a huge productivity gain because I think they have sort of the skillset and the experience to evaluate the output. Whereas if you put a similar task in the hands of a junior developer, maybe it's not the same. And that's because of the skill gap.