The impact of AI-based tools on embedded business lines

The embedded world is a dynamic landscape where development teams, driven by the rapid pace of technology and constrained by the demands of the environment, face numerous challenges and opportunities. 

To understand them, we approached several players in a sector that goes beyond software. The experts we spoke to come from a wide range of sectors: banking, medical, industrial, automotive, defense, robotics… and have a vision of embedded systems that is both common and unique to each sector. 

The major challenges facing the embedded world is a detailed qualitative analysis of interviews with players in the embedded world. We discuss key trends and specific issues that have emerged, allowing us to visualize the future of this field. 

A topic that takes us right to the heart of embedded trends and allows companies to express their views on how to overcome the obstacles and challenges posed by this sector. A previous article, The biggest perceived constraints in embedded projects kicked off the topic of the rise of Artificial Intelligence (AI) and AI-based tools, which are taking up more and more space and challenging embedded professionals. 

The advent of AI and the impact of digital technology on the environment are among the hot topics of the moment, raising many questions about the use and integration of AI-based tools in embedded projects. To address this topic, we submitted the following question during our exchanges with players in the embedded world: How do you perceive the arrival of AI and AI-based tools in the embedded business lines?

The arrival and development of AI in the embedded sector is not seen as a positive step forward by the majority of respondents. 47% of respondents do not use AI-based tools and even refuse to consider using them. One of the most frequently mentioned reasons for this is a lack of information and insight into the usefulness of these AI-based tools. Respondents report that the recommendations provided by these tools are not clear or precise enough to be used with confidence. To take it a step further, some respondents said they had no idea “what is behind these tools”. One thing is for sure: some of the more resistant opinions are categorical. 

It should be noted, however, that this conclusion from the various exchanges is based on the respondents’ testing of AI-based tools. Whenever possible, or when developments allowed, most players tested these tools on embedded projects. For example, ChatGPT was tested and used to write a classic algorithm. Conclusion: the tool couldn’t do it. Overall, the results were never conclusive (or even disastrous for some). It was also used to generate unit tests. Although the tool was able to generate working code, verification by the developer was required. These results show that there is a real need for human expertise in the embedded sector. Certain skills cannot be replaced by AI, especially in critical embedded projects such as those in the aerospace sector, where trajectory optimization is a high-impact challenge. 

Embedded projects also face numerous standards constraints, such as DO160, which qualifies equipment used in the aerospace sector and imposes limits on the use of AI tools. Some of the players interviewed expressed a lack of visibility when it comes to AI standards compliance.

The energy consumption aspect of AI-based tools is also raised by respondents. We find that embedded players have little confidence in AI that consumes a lot of resources, or even too much. Today, AI-based tools consume ten times as much energy as more frugal tools for the same request.

While some players in the embedded world remain categorically opposed to the use of AI, others support its use and even proclaim the obligation to use these tools in order to remain competitive in the market (11% of respondents described the use of AI as “mandatory”). AI can therefore also be seen as a competitive advantage, allowing any player to stand out from the crowd.

A total of 23% of respondents use AI-based tools. This is in response to a major problem: software development time. Development teams are incorporating AI into their processes to save time, but not only that: once development is complete, AI is needed again for code fixes. 

In some organizations, the use of AI by developers may need to be overseen by management. In the cases cited in this study, although there was no specific approval for the use of AI, it was decided that if AI-based tools enable developers to make their work easier, then it is welcome. 

AI can be seen as an additional tool that can speed up development and allow people to focus on other, higher value-added tasks. It is not seen as a replacement, but as a complement to existing skills, potentially optimizing code and exploring new avenues of development.

While they do not have a clear opinion on the subject, 29% of players surveyed do not use AI-based tools, but are interested in using them. 

For some, the need is already there, and they are even willing to take a training course on the subject. While some of the respondents do not currently use AI-based tools internally, they see their use as potentially interesting at the software design level, where it would improve development cycles and differentiate the way we work.

For others, the need will emerge in the medium to long term. The interest is there, but the milestone has not yet been reached because the integration of AI-based tools in the embedded domain is too complex and certification is not possible.

There’s also a lot of interest from the new generation of developers who are integrating AI into their work, using ChatGPT to learn C and C++ programming languages, for example, and to develop new features. What about this new influence? For companies that are more concerned about the issues raised by AI, the question is how to train developers to use AI tools while encouraging them to think critically.

Interested in contributing your comments to this study?

LET’S TALK!

The impact of AI tools on embedded business lines is an article in the series Major issues in the embedded world. It is the result of a qualitative analysis carried out with players in the embedded sector. We would like to thank them for sharing their views on the specific trends and issues facing the embedded world.