Review article: enhancing the power of artificial intelligence in mechanical design

. Artificial intelligence-powered tools are emerging lately due to the improvement of computational power and advancements in other fields. This technological progress has led to the emergence of many tools that are able to perform tasks, and even outperform humans, leading to the automation of many processes. However, this also comes with certain limitations. This work will explore generative design and its primarily application in mechanical engineering. The study will present how generative design is reshaping mechanical engineering by optimizing structures and discussing the limitations of this technique


Introduction
Progress in modern times is moving faster than one can keep up with. New technologies are introduced every day, and their integration and dependence are increasing. It has become easier to use intuitive and user-friendly tools. Technological progress and the integration of new technologies in various industries are driving improvements in different fields. Therefore, the correct utilization of technology will benefit the respective industry [1].
In the past, everything was tangible and guided by humans. However, new innovations are now more focused on software improvements, which enhance the functionality and quality of hardware and mechanics [2]. Industries are thriving on modern AI-powered tools (AI stands for artificial intelligence). The implementation of AI has contributed to significant improvements in various aspects of industries. AI-powered tools are widely utilized to increase production rates, reduce waste, streamline workforce, enhance precision, improve the quality of the final product, and more.
This work, however, will focus on the usage of this technology in the field of mechanical engineering, specifically in the domain of generative design, which aims to simplify mechanical design.
Traditionally, mechanical design is done using software that is primarily operated manually. To build any model in this software, one needs to use various provided tools within the interface. Similarly, the simulation of the given object requires different manipulations and takes some time to perform [3]. Each modification necessitates new steps to achieve the desired results. However, with the introduction of AI into these software systems, they have entered a new era of modeling. This integration allows for a more intuitive approach to modeling, modifying, simulating, and other improvements that will be outlined in this work.

Problems and solutions
The modern world does not remain stagnant, and technological progress is endless. While some issues are addressed through the use of new technologies, others remain undisclosed or new ones emerge on a daily basis. These challenges require quick and innovative solutions. We have reached a point where the human brain alone cannot keep up with the rapidly changing world. It is widely recognized that time and cost are more valuable than ever, leading to the emergence of new tools to meet these needs. One promising and novel solution to these issues is artificial intelligence (AI) and the tools that harness its potential. Among these techniques, generative design, powered by AI, is revolutionizing the field of modeling.

Generative design
AI has made its way into almost every industry and field, and three-dimensional object modeling is no exception, although it took longer than other domains like photos, videos, and text. This technique is used to optimize design-related issues. Generative design utilizes a pre-built model to generate more optimal solutions, taking into account a set of parameters and constraints at each iteration. This iterative process results in generating different versions of the same object, but with improvements in terms of ergonomic design [4].
Generative design aims to mimic nature, as it tends to provide the most sophisticated results when given control. The technique enhances the overall required parameters by eliminating weaknesses and continuously iterating for improvement, guided by each individual case. It's worth noting that the technique may produce outcomes that are not always feasible for humans (designers). However, this process of improvement consistently leads to the most ergonomic and unconventional design solutions.
This process differs from traditional methods by involving AI in half of the work. Designers and AI tools can collaborate to provide a solid outcome that best suits the intended purpose. Boundary conditions such as objectives, parameters, and constraints are set to achieve the desired goal. In reality, generative design methods are designed to shift from the old approach where design processes were driven by humans. By offloading certain complex steps to AI, designers have more time to focus on their creative abilities and produce their best work [5,6].

Merge of mechanical engineering and generative design
Mechanical engineering design is similar to any form of modelling, and the use of generative design (GD) can enhance its advantages. The application of GD in mechanical engineering can expand the capabilities of traditional methods and yield better outcomes for all given specifications. This paragraph will cover some of its uses in mechanical engineering. The following applications will be discussed:

Shape modification
Generative design can be applied to any shape in mechanical engineering. The results may differ from what humans typically create and perceive, as they are optimized based on given criteria. This technique allows scientists to undergo a complex and time-consuming iterative process to reshape objects in the most optimized manner [7,8].

Optimization in terms of weight and strength
The robustness of a part against external and internal forces is a crucial criterion in mechanical engineering. Material usage also plays a significant role in product cost. Therefore, the use of generative design in this field is essential. It can easily modify the structure of an object to meet strength and safety requirements for its intended environment. By optimizing the structure, material usage is minimized, resulting in lighter products while maintaining structural integrity under various environmental conditions. Generative design also improves other structural aspects.

Additive manufacturing and generative design
Additive manufacturing, commonly known as 3D printing, involves building parts layer by layer. Different 3D printers are used for various purposes, but quality approaches are always important. Generative design is employed to create precise and intricate support structures, as well as different lattice structures. The technique utilizes a 3D model and applies iterative processes to minimize excessive material usage while preserving overall structural integrity and ergonomic considerations.

Design variations
Generative design can generate different designs for a particular part through an iterative process. Having various design variations provides greater control and diversity. Different boundary conditions can be applied to observe the behaviour of the structure under varying conditions and choose the desired outcome.

Simulation and integration of generative design
Simulation plays a vital role in engineering, not only for validating results but also providing different data. Traditional methods of simulation can be time-consuming, whereas AIpowered tools significantly reduce the time required. For example, simulation can identify areas where a certain load would cause failure, and this data is fed back to the AI, which determines the need for additional material in those areas to handle stress. Alternatively, the technique can optimize material distribution while ensuring the robustness against applied constraints using various methods mentioned earlier.

Generative design in general
It is evident that generative design is a valuable tool, not only for creating simple 3D objects but also for real-life applications in mechanical engineering, resulting in unconventional yet optimized structures. This technique offers numerous useful functionalities for engineering, but like any technology, it has its advantages and limitations in specific areas.

limitations of generative design in mechanical engineering
This section will cover some limitations of the generative design technique as it currently stands. While it provides solid results, it is important to acknowledge that all technological solutions have their advantages and limitations. The following areas will be addressed to highlight the need for improvement in generative design to overcome present and future limitations [9,10].

Appearance of the final product
The first disadvantage is primarily related to aesthetics rather than performance. As mentioned before, the optimization process can result in unconventional designs that may not appeal to some customers. The complexity of the final product's structure can also make manufacturing, cost efficiency, and replication challenging.

Lack of multidirectional approach
Generative design primarily focuses on structural efficiency, which may overlook other important parameters. While it leads to optimal outcomes in terms of force dissipation and mechanical properties, it can introduce challenges in terms of assembly, interaction with other parts, maintenance, and other considerations. Decisions need to be made regarding which parameters to prioritize and which to ignore.

The need for precise input values
The technique heavily relies on accurate input values to generate optimal results. Even slight changes in the input can lead to different outcomes, necessitating a thorough understanding of how to provide input data for the best results. Additionally, the technique often suggests using a specific material, limiting potential variations.

Ethical issues
Introducing AI in mechanical engineering raises ethical concerns, as it may replace and eliminate certain jobs and professions, affecting employment opportunities. It can also limit the need for continuous learning as AI takes over tasks traditionally performed by humans.
By recognizing these limitations, it becomes clear that generative design is still in its early stages of development. Several issues need to be addressed to reduce limitations and provide more flexibility in utilizing its functionalities.

Future of generative design
As we move towards the widespread use of artificial intelligence in various areas, the future of generative design appears promising in terms of performance. Currently, AI is already being utilized for modelling purposes through specific requests, indicating that AI will not only be involved in post-processing but also in providing parameters from the beginning. This means that the refinement process will cover multiple criteria, including not only structural improvements but also considerations such as overall appearance, assembly, and maintenance. As a result, the aforementioned limitations can be overcome, with the exception of the issue of employment. The advancement of this technique will directly impact job loss for individuals and the closure of certain companies. Furthermore, it is possible that the tool will guide researchers to follow more efficient paths to achieve their overall objectives. By foreseeing future issues that humans may not be able to predict, the tool can provide the best possible outcomes.

Conclusion
In conclusion, generative design needs to undergo improvements in various areas to fully unleash its potential and overcome limitations that currently hinder its widespread adoption.