Under the premise of increasing productivity and precision, artificial intelligence is currently making history in the factories of the manufacturing industry: predictive maintenance based on machine learning supports the longevity of systems; AI-controlled robotics, model simulations such as digital twins and automation accelerate production; and quality control relies on AI-supported image and error detection to identify and avoid defects. Behind all these processes are teams of developers who are themselves using more and more generative AI tools to drive application development forward.

Away from the hype

However, if companies are not blinded by the potential of AI, pitfalls can also be identified behind the technology - especially when it comes to the issue of trust. In order to be able to use AI in the development of industrial applications, companies must be able to rely on the generated code being secure and performing correctly. This requires a precise review of the code, which is still manageable for small quantities, but can quickly get out of hand given the - still existing - unpredictability of generative AI. Depending on the context and request, these processes vary greatly and are therefore difficult to predict.

As an iterative process, the development of software is fundamentally characterized by constant adjustments, which is why companies must expect a certain degree of uncertainty. However, the continuous production of code snippets by AI makes it difficult for employees to understand the code in its entirety. Ultimately, this leads to a confusing accumulation of code, which inevitably leads to technical debt.

An underestimated problem

Technical debt is an often underestimated problem and - since Ward Cunningham coined the term in 1992 - has often been used as an excuse for releasing fast and bad code in the interest of development speed. It refers to the technologies, time and money that companies spend on maintaining outdated, unworkable and buggy code instead of developing new ideas. In order to achieve top performance, technical debts are often factored in. However, it is important that they are paid off before they trigger a whole series of problems. If there is already a large backlog of technical debt at the beginning of a development cycle, it is often not possible to react quickly to new opportunities and challenges. In the long term, these debts lead to overloaded software and a limited ability to compete and innovate.

AI and technical debt

In the context of AI, too, fast solutions currently seem to be the best option for riding the wave. But AI doesn't just speed up the development process. Companies also have to deal with issues of security, governance, code quality and application lifecycle management more quickly and under greater time pressure. If generative AI is only used to create huge amounts of code without considering where it goes or who is responsible for its maintenance, it could exacerbate an organization's existing technical debt. Instead of accelerating innovation, AI would actually slow it down in the long term. Companies should therefore always be aware of technical debt. Dealing with it requires a balance between quality and speed - quality to provide an engaging experience for users and speed to achieve business goals. Technical debt may seem harmless. However, if it gets out of control, speed and agility are simply no longer an option.

How technical debt can be curbed

Christoph Volkmer is Regional Vice President EMEA Central Outsystems

© Outsystems

The use of AI to produce code quickly and therefore produce goods more efficiently may be tempting for the industry. However, it would be even more efficient to invest in technologies that enable AI to be used in a way that your own teams can understand, oversee and trust. AI-supported low-code platforms that visualize changes to applications in real time, making testing, staging and monitoring much more efficient, will play an important role in this. It is important for the industry to use AI responsibly and sustainably. This also includes a strategic rethink, specifically the consideration of opting for high-quality solutions.