Reskilling for AI

It used to be a common refrain from clients: “We have a complex problem, so we want your best brains on it.” It’s no surprise: given the complexity of today’s enterprise technology estates, years of knowhow can accelerate the time to value. But recently, we’ve had a different kind of request from one client. They wanted more graduates to be assigned for their projects.

This isn’t about a client looking to trim back expenses. It’s all down to AI and the perception that the younger generation are better able to leverage AI capabilities to deliver value. Indeed, by some estimates, two-thirds of enterprise leaders now say they wouldn’t hire a candidate without AI skills.

Talent in the AI generation

A grad-heavy development team is an intriguing idea. But is it realistic? How should you think about talent in the age of AI? What are the skills that will position your organization best? The answers, it transpires, aren’t always the obvious ones. 

Firstly, it’s worth exploring the assumption that younger developers will be better at using AI tools. It’s probably fair to say that the more experienced you are, the harder it can be to adapt to new ways of working. One of the benefits of experience is the workflows and processes that enable you to reach your goals faster become like muscle memory. The downside is that embracing new ideas, such as integrating AI tools into your work, requires intentional effort. The learning curve can feel steep. This highlights a potential generational difference, where younger, more "AI-native" individuals may find these tools more intuitive as they begin their careers.

But it doesn’t necessarily mean that deploying a band of graduate developers, all equipped with the latest AI tools, will accelerate your time to value.

Consider the example of AI-enabled software development. We’ve seen firsthand how AI coding assistants such as Cursor can generate code at eye-catching speeds; Cursor can even detect compilation errors in generated code and proactively correct them. But when building enterprise-grade software, functional correctness is just the starting point. We need code that’s safe, maintainable and makes reasonable use of resources such as compute power. A focus on just using coding assistants might result in short-term gains but risks potential long-term pain related to code quality and maintainability.

A balanced approached to talent

Here we see the pitfalls of the over-reliance on graduate-packed teams. While graduates might be more comfortable with new AI tools, they often lack the experience to discern "what good looks like." Experienced technologists bring "battle wounds" and the ability to anticipate long-term impacts and quality concerns associated with AI-generated code. They possess the critical ability to look ahead and understand whether the generated output truly meets business goals.

The optimal approach lies in pairing experienced individuals with those earlier in their careers. This combination can be incredibly complementary. Less experienced team members often ask challenging questions and readily adopt new technologies, while seasoned professionals provide crucial context, quality oversight and the ability to foresee potential pitfalls. If there’s a willingness to learn from each other, and an understanding that different perspectives can be useful, you teams are more likely to be able to truly explore and exploit AI's potential. Maintaining a good mix of experience levels will continue to be the norm, and the need for mid-level technologists remains significant as they develop the crucial experience of "knowing what good looks like." This progression is reassuring for mid-level professionals, as their developed experience in asking the right questions and anticipating different scenarios remains highly valuable.

Skills like testing and clearly defining requirements become more crucial than ever. Furthermore, the fundamentals of good development practices, such as code reviews and pair programming, are increasingly important in an AI-assisted environment to ensure quality and prevent the unchecked deployment of potentially flawed AI outputs. 

It feels likely that AI isn’t going to radically change team structures, nonetheless, you can bet it will change the skills companies are looking for. According to industry analysts, the rise of AI will push engineers toward an “AI-first mindset,” where they focus on guiding AI and providing it context. In this future, skills like natural-language prompt engineering and building RAG pipelines become part of the standard toolkit.

Of course, software engineers have always needed to think about the skill sets that are in demand. But the rise of AI brings sharp focus on another question: are there a set of skills that enable people to effectively utilize AI tools? Here, a fascinating observation emerges: the interaction with AI is akin to the Socratic method. To elicit useful responses, you have to ask open-ended and well-crafted questions. Otherwise, the AI might simply confirm existing biases. This necessitates critical thinking — the ability to analyze the AI's output and assess its validity and potential weaknesses within a specific context. You want your teams to include ethical considerations and responsible AI usage in their thinking, so that engineers understand the broader implications of deploying AI systems, including issues like bias and data privacy.

You can see how this need for higher-level skills plays out in something like prompt engineering. It’s a skill that is highly sought after, but the fundamentals to prompt engineering aren’t about technical literacy, it’s about the ability to ask the right questions.

The growing emphasis of skills such as on critical thinking, questioning and context awareness raises a tantalizing idea: such skills aren’t just the preserve of computer scientists, could companies broaden their search for talent beyond traditional computer science graduates? That may go some way to address long-standing issues of representation in the tech industry. By valuing these "higher-order thinking tasks" that AI is not yet adept at, the industry can attract individuals with valuable skills learned in diverse educational fields, potentially accelerating progress toward a more inclusive workforce.

It’s already crystal clear that AI is changing the way we think about developers and the skills they need. While AI tools offer significant potential for productivity and innovation, they also necessitate a fundamental shift in how organizations think about talent and skills. The future of tech talent lies not in replacing human skills with AI, but in effectively combining foundational expertise with AI-enhanced capabilities to meet the rapidly evolving enterprise requirements.