TL;DR
AI can process ISO-GPS data, make it available with pinpoint accuracy and support specialist users, for example when interpreting drawings. The more autonomously it operates, the more important the ‘human in the loop’ remains: AI provides support – humans make the decisions.
Collaboration between Artificial Intelligence and ISO GPS
Artificial intelligence is currently developing at a rapid pace. Many applications that were still a distant prospect just a few years ago are already available today. Consequently, the question is increasingly being asked in the technical sector – and particularly at ISO GPS – where AI can provide useful support, and where its limitations lie.
First and foremost, it is important not to treat every form of AI as the same. There are several stages of development between a simple chatbot and a system that operates largely autonomously.
At the initial stage, the AI simply responds to questions. For example, it can explain terms, summarise the content of standards in an accessible way, or clarify what a particular symbol on a technical drawing means. In this role, it primarily serves as a knowledge assistant.
At a second stage, the AI is already able to work more effectively with specific information. For example, it could analyse a drawing, recognise tolerance frames, reference symbols or dimensional values, and explain to the user which pieces of information are related. The AI thus helps not only with searching for knowledge, but also with interpreting existing information.
A further step involves the AI making suggestions. When it comes to tolerance, for example, it could highlight which specifications might be relevant to a particular task or which points a designer should still check. It thus takes on a sort of co-pilot role: it processes information and assists with decision-making.
So-called agent-based systems go even further. They could carry out individual work steps independently, for example analysing a CAD model, checking existing PMI information, taking company rules into account or preparing proposals for tolerance setting. Humans would then no longer carry out every single step themselves, but would instead review and approve certain decisions.
Ultimately, the highest level would be a largely autonomous AI that independently generates a complete set of tolerances. It is conceivable, for example, that a system might be given a 3D model and functional requirements and, based on these, independently create a complete ISO-GPS tolerance concept. Technologically speaking, this is an exciting vision. At the same time, it remains to be seen whether such complete automation is technically reliable, economically viable and even desirable.
Can an agent tolerate geometry?
So-called agent-based systems go even further. They could carry out individual work steps independently, for example analysing a CAD model, checking existing PMI information, taking company rules into account or preparing proposals for tolerance setting. Humans would then no longer carry out every single step themselves, but would instead review and approve certain decisions.
Ultimately, the highest level would be a largely autonomous AI that independently generates a complete set of tolerances. It is conceivable, for example, that a system might be given a 3D model and functional requirements and, based on these, independently create a complete ISO-GPS tolerance concept. Technologically speaking, this is an exciting vision. At the same time, it remains to be seen whether such complete automation is technically reliable, economically viable and even desirable.
Current limitations and the importance of the user
The ISO-GPS system, in particular, highlights a key limitation of artificial intelligence. Technically correct tolerance determination does not arise solely from knowledge of standards. The designer is familiar with the component’s function, experience gained from similar products, manufacturing processes, assembly conditions, measurement capabilities and economic constraints. Much of this information is not fully documented and cannot be readily recognised by an AI.
That is why the so-called „human-in-the-loop“ approach remains crucial. This means that humans remain part of the decision-making process. AI can find, sort and process information, and make suggestions. For example, it can help to make complex ISO-GPS rules easier to understand or identify relevant details on a drawing more quickly. However, the technical assessment and responsibility should remain with the specialist user.
How can AI support the ISO GPS application?
From today’s perspective, this is precisely where there lies great potential. The benefit of AI need not lie in replacing the design engineer or measurement technician. What is far more interesting is the Fra0+
In practical terms, this means that the focus today should not be on a fully autonomous „AI designer“. A much more realistic approach is to develop intelligent assistance systems that help specialist users to find, understand and apply ISO-GPS knowledge.
The crucial question, therefore, is not so much: „Will AI be capable of exercising tolerance in the future?“ but rather: „How can we use AI today to enable people to make better, faster and more transparent technical decisions?“
What does GPSlife do?
GPSlife is already working on new approaches to effectively combine artificial intelligence with ISO GPS knowledge and technical applications. Our learning materials already enable interaction with a chatbot that supports learners with any questions they may have directly during the learning process. In addition, we are developing further concepts for intelligent assistance systems centred on ISO GPS. Anyone interested in these developments or wishing to discuss them is welcome to get in touch with us.
Please feel free to use our Contact form!
You can subscribe to GPSlife’s free ISO GPS newsletter here to stay up to date.
