Peter Guagenti is president at Tabnine. Peter is an accomplished entrepreneur, and has been working in AI business tools for 10+ years.
Software development just isn’t what it used to be. The role of the developer is more challenging today than it was before the pandemic, and it’s exponentially more difficult than at the start of the mobile era 10-plus years ago.
Why? The world’s seemingly insatiable appetite for software. To put it into numbers, Google Play adds about 1,270 new applications per day, and the average enterprise has over 1,000 applications in use at any given time. The burden of creating, maintaining and modernizing all of this code is only increasing.
The combination of technological advancements and fierce competition between businesses has created the need to build more applications faster. Devs are tasked with creating and maintaining these apps—including contending with their increasing complexity—while simultaneously dealing with a growing pool of technical debt that has accumulated over time.
The final piece of the puzzle contributing to this challenging new environment is a talent shortage. According to one study, “the shortage of developers in the US will exceed 1.2 million.” This talent shortage is so severe that “the US economy [is] at risk of an unrealized [GDP] output of $162 billion.” Developers are under pressure to deliver a higher output than ever before, and many of them are doing the work of several software developers.
AI to the rescue—or just more risk?
Generative AI-powered solutions have emerged as a promising path forward to help developers work more efficiently (and therefore keep up with demand) and reduce technical debt. The technology has already demonstrated its value for automating many of devs’ repetitive tasks (think writing new code, maintaining existing code, writing tests, etc.), in turn freeing them up to focus on more creative, high-value work.
However, there currently are several impediments to the effective adoption of AI. Three top concerns for CIOs include:
1. Resistance To Change: Humans are inherently scared of—and therefore really resistant to—change. Automation always represents a major change to business processes and a threat to how people do their jobs. Engineering teams and IT professionals are likely to see AI as a threat to how they currently and comfortably work and are fearful of how they might be expected to work in the future.
2. Privacy And Compliance Concerns: Privacy is typically the number one concern among CIOs when it comes to adopting generative AI, and they’re not wrong to worry. Right now, big tech companies are talking out of both sides of their mouths, claiming that they protect privacy and honor copyright and IP while demonstrably not behaving that way.
How is a company to know for absolute certain that their protected data is not being captured and leveraged in a way they don’t intend? In addition, with AI models pulling data from all corners of the digital world, the potential for generative AI to produce output that can be traced back to copyrighted or restricted works is high, so it is no wonder that CIOs are demonstrating extreme caution.
3. Technical Limitations: One of the biggest hurdles to organizations’ embrace of AI is limitations in the AI technology itself—or, more to the point, organizations’ limited understanding of the technology’s limitations. Generative AI has been hyped as a replacement for the works they are creating, and that hype has applied to the perception of AI coding assistants as well. But the reality is more nuanced than that. The current state of generative AI tools offers massive productivity gains for many tasks but are best wielded by experts who know their craft. Knowing where and how to deploy AI tools to maximum effect is critical.
These hurdles are admittedly high, but to unlock the productivity gains we need, organizations must find a way to overcome them. Engineering teams need to transform or be transformed.
How? Here are three ways to smooth the path to AI adoption.
1. Understand how a vendor’s terms of service may impact business.
It’s critical to review the terms of service of the platforms you consider (How will they use your data? What information is retained?) and to understand how they trained their code (is the code they trained on licensed for use by your company?). You do not need to sacrifice privacy, control and compliance to adopt AI tools. There is enough diversity in the available offerings that you can find a solution that will both delight your developers and eliminate the concerns of your legal counsel.
2. Understand the landscape and be unafraid to experiment.
It will also be critical to develop a culture of understanding and ongoing education around AI and to look for tools that will help developers and engineering teams safely, effectively and purposely harness the power and potential of AI.
For example, generative AI-powered coding assistants like Github Copilot and AWS CodeWhisperer can boost productivity and help developers create, test, document and fix code faster. AI-enabled chat agents contribute to developer knowledge and serve as hands-on coaches for both new developers and experienced developers needing a refresher on specific languages, frameworks or techniques. Multiple independent studies have shown consistent, double-digit productivity gains across the average developer’s total workload and upwards of 50% automation in software generation itself.
3. View AI as a way to supercharge current strengths.
Generative AI isn’t a “set it and forget it” solution or a perfect replacement for a developer’s day-to-day tasks (at least not yet). Despite the hype around AI, these tools remain assistants and recommenders, not replacements for the full capability of a skilled software engineer.
To benefit most from AI software development tools, engineering teams should consider them less like autonomous machines and more like Iron Man suits for the mind. Expect to make subtle but important changes in your approach to accommodate asking for guidance, autonomously generating code and content and reviewing AI recommendations. You will be much faster and more efficient for it.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?