• Written by Azhar Aliyar, Director

A research and development team leader checks the project management system to see which tasks are pending and which have been completed, but the AI system they are using to support their work is not giving them the answers they need because it lacks access to certain critical systems.

The systems that support AI in research and development

The AI system in a research and development deployment depends on data from several specific systems, including the test results and failure analyses database, which contains detailed information about the performance of different components and materials. This data is essential for identifying patterns and trends that can inform design decisions. The design review records system is another critical source of data, as it contains information about the evolution of the design and the decisions that were made at each stage.

The timesheets and allocation system is also an important source of data, as it contains information about how much time and resources have been devoted to different tasks and projects. This data can be used to identify areas where the team is struggling and where additional support may be needed. Supplier assessments are another key area of data, as they contain information about the performance and reliability of different suppliers and can be used to inform procurement decisions.

For example, a research and development engineer working on a new product design can use the AI system to analyze data from the test results and failure analyses database to identify potential weaknesses in the design. They can then use this information to inform their design decisions and reduce the risk of costly rework or product failures.

What happens when a system is missing

If one of the critical systems that supports the AI is missing or not integrated, the AI system will not be able to provide accurate or complete answers. For instance, if the test results and failure analyses database is not available, the AI system will not be able to identify patterns and trends in the data, and the research and development team will not be able to make informed decisions about their design. This can lead to delays, cost overruns, and reduced product quality.

A missing system can also lead to biased or incomplete results, as the AI system will be working with incomplete or inaccurate data. This can be particularly problematic in research and development, where small errors or biases can have significant consequences. For example, a biased AI system may recommend a design change that seems optimal based on the available data, but that actually introduces new problems or weaknesses.

Defining terms and concepts

Before deploying an AI system in a research and development context, it is essential to define key terms and concepts, such as what constitutes a "test result" or a "failure analysis". This may seem straightforward, but in practice, different teams and stakeholders may have different understandings of these terms, and the AI system will need to be programmed to recognize and work with these definitions.

For example, a team may define a "test result" as a specific type of data that is generated by a particular type of test equipment. The AI system will need to be programmed to recognize and work with this definition, and to distinguish it from other types of data that may be generated by different equipment or processes. This requires careful planning and coordination between the research and development team, the AI system developers, and other stakeholders.

More information about research and development can be found on our research and development page.

Quality problems that surface later

Even if the AI system is initially successful, quality problems can still surface later if the underlying data is not accurate or complete. For instance, if the supplier assessments are not up to date or accurate, the AI system may recommend a supplier that is no longer reliable or that has changed its manufacturing processes. This can lead to delays, cost overruns, and reduced product quality.

Quality problems can also arise if the AI system is not properly maintained or updated. For example, if the test results and failure analyses database is not regularly updated, the AI system may not be able to identify new patterns or trends in the data, and the research and development team may not be able to make informed decisions about their design.

When AI is not the right choice

While AI can be a powerful tool in research and development, there are certain situations where it may not be the right choice. For example, if the research and development team is working on a highly complex or novel project, the AI system may not have enough data or context to provide accurate or useful results. In these cases, more traditional research and development methods may be more effective.

If you are considering deploying an AI system in a research and development context, but are unsure about whether it is the right choice, you can contact us for more information and guidance.