Turbocharging software with AI: Fast, not furious

As organizations increasingly embrace generative AI in software engineering, both transformative opportunities and significant risks are coming to light. Our latest research provides concrete guidance on how organizations can safely harness Gen AI’s full potential while transforming their software engineering capabilities.

What's everyone saying?

• Gen AI is widely believed to be on the verge of driving value creation and productivity gains for organizations that embrace it at scale, potentially ending decades of stagnating productivity. Experts compare its transformative power to that of electric motors or personal computers.

What do we have to say?

• Gen AI adoption is still in its early stages, with only 10% of organizations currently using it in software engineering, mainly for coding and UX tasks, finds our recent study, Turbocharging software with Gen AI, which surveyed 1,097 senior executives and 1,092 software professionals. However, momentum is building fast, as evidenced by the figures below.

• This enthusiasm spurs from the significant time savings Gen AI delivers in crafting documentation, writing code and scripts, debugging and testing, and project management. While the focus often centers on coding, Gen AI has applications throughout the software development lifecycle, including business requirements, UX design, maintenance, testing, and collaboration.
• For organizations already applying Gen AI to software engineering, productivity gains range from 7 to 18%, depending on the industry, company size, and tools employed. But while these are significant gains, productivity shouldn’t be the main motive for investing in Gen AI. Instead, it’s about how organizations reinvest the time saved and transform their culture and mindsets. The key question isn’t “when will we see this productivity boost?”, but “how?”
• In fact, it’s less about investing in Gen AI for its own sake, and more about mastering software engineering, elevated by AI. Many organizations recognize this and are channeling these productivity gains into innovation. Primarily, they plan to use the time freed up by Gen AI to focus software professionals’ efforts on developing new features, products, and services.
• Investing in people by upskilling software professionals, in both technical and broader business domains, is the next most common use of these productivity gains. Organizations are also reallocating software professionals to more complex tasks that require human perspective and deeper understanding, while assigning routine tasks to Gen AI.
• In short, Gen AI is prompting organizations to invest in talent. Its widespread adoption should, therefore, be an exciting prospect for workers. In fact, a majority of Gen AI users are already reporting increased job satisfaction.

• However, unchecked use of Gen AI can also expose organizations to multiple threats, especially in the early days we are currently living. The primary factors? First, a lack of appropriate tools: Less than a third of organizations have the platforms, tools, and talent prerequisites in place to implement Gen AI for software engineering. Over 60% lack a governance framework and upskilling programs for Gen AI for software, while nearly a third of the workforce is self-training. And 63% use unauthorized AI tools and solutions. These practices expose organizations to functional errors and risks (AI hallucinations, bias, etc.), security breaches, and legal issues (privacy, code leakage, intellectual property (IP) infringements, etc.). These risks highlight the urgent need for proper governance and oversight.

Who's doing it right?

• At the Paris Motor Show in October, Jean-Vianney Chiron , Head of AI Transformation at 米其林 , shared insights on how the French engineering materials manufacturer, an early adopter of Gen AI, has used it to support its software development cycle. Among the key insights? Generative AI can help retain the best talents; in fact, developers are actively requesting it, and those that have already had a taste of its impact don't want to miss out on it anymore.

What's the bottom line?

The good news is that organizations can take concrete actions to harness the full potential of Gen AI for software engineering by mitigating security risks, carefully selecting use cases, and putting people at the heart of this transformation.

1. Organizations need to choose and prioritize high benefit use cases. These can include crafting literature and documentation, coding assistance, debugging and testing, identifying security vulnerabilities, modernizing code (through translation, migration, or conversion), or preparing and evaluating RFPs.
2. To mitigate risks around security, IP, or copyright issues and code leakage, organizations should adopt a thorough risk management approach. This means conducting a risk assessment for each selected use case and involving internal legal, IP, cybersecurity, and data protection experts from the earliest stages. More generally, organizations need to proceed with caution and ensure human oversight is in place.
3. To ensure optimal use of Gen AI, companies should transform their software organization, rethinking established processes and developing new capabilities. This might involve learning how to pair with AI to analyze output and iterate it until reaching an acceptable solution, developing and updating prompt libraries and playbooks for software teams, or addressing complex bugs that AI can’t solve or prevent.
4. Organizations can prepare for Gen AI use by delivering technology prerequisites. This means building a repository of platforms and tools for a seamless and augmented software engineering experience, as well as privately and safely contextualizing Gen AI assistants with an organization’s own content.
5. As of today, 48% of organizations have no standard metrics to measure the success of generative AI use in software. Organizations should adopt a measurement protocol for Gen AI involving a measurement approach, use case prioritization, metrics, team, and a well-defined process to create actionable and reliable results.
6. Finally, people must remain at the heart of this transformation. This means organizations will have to create a learning culture, which can be achieved by providing upskilling and cross-skilling opportunities.

Looking for more?

• Head here to learn how to integrate Gen AI across the full software lifecycle.
• Looking for a glimpse into what the future holds for AI in software engineering? We’ve got you covered.
• On a recent episode, the hosts of our Cloud Realities podcast sat down with AI business transformation expert Nicole Onuta about the growing impact of Gen AI on the software development lifecycle and how hyper-automation is fast becoming a practical reality. Tune in!

And you, what are you saying?

We’d love to hear about how Gen AI is transforming software engineering in your industry and your organization’s experience in the matter, whether you’re just starting to explore its potential or already seeing results. Head to the comments section below to join the discussion.