Last week, I attended Autodesk’s 2020 product conference. The opening ceremony was Autodesk researcher Zheng Qingyi presenting several simple and straightforward cases of AI in design software.
AI has recently been a hot topic. The question, “how many professions will be replaced by AI?” is frequently asked. Some speculations range from blue-collar workers to lawyers. Shortly after the release of Autodraw, it was believed that creative professionals like graphic designers could be replaced as well.
Those who have used Rhino Grasshopper should be familiar with this topic. Rendering software can quickly produce numerous designs, and the designer only needs to set or adjust conditions. Designers haven’t been dismissed but have become the “supervisor”. The designer learns “management rules” (i.e. a programming language), gives “direction” (sets parameters), and finally “evaluates and adjusts”. This is an ideal human-machine collaboration.
The recent emergence of DeepFake technology, a machine-learning algorithm to analyze libraries of audio and video of an individual, is even more controversial. Once analyzed, DeepFake data can be combined to synthesize film of the same subject doing and saying pretty much anything. One famous DeepFake is U.S. president Obama making a public statement on the dangers of DeepFakes.
Photoshop’s new “content-aware fill” in 2019 CC allows users to remove objects from an image, or to elongate or change the proportion of objects with ease. Adobe also introduced the “Face Recognition & Liquify” tool in Photoshop. The software automatically maps facial features and provides sliders to manipulate them. Recently AI has been able to detect facial manipulations with 99% accuracy, and even reverse the manipulations made by other software.
Qingyi explained three stages of AI’s role in design software, similar to the five stages of autonomous driving:
Design Automation (For you)
This is similar to the Rhino+Grasshopper generative designs. Currently, in the field of architecture, jewelry, and digital design, this technology is very mature. NASA collaborated with Autodesk to design the Lander concept. The Lander design proved difficult due to its weight requirements and its potential to withstand a variety of difficult environmental conditions. After one and a half months, a spider-like lander design was produced that weighed 35% less than the original design and built with 3D printed aluminum components.
(The NASA Autodesk collaboration on the Lander project.)
Design Interaction (With you)
Further integration of AI allows for an autocompletion of what is being drawn. For example, when drawing an interior design scene, AI can auto-complete doorways or windows. You can start to draw simple shapes such as circles or squares and those will also be autocompleted. Generative design combines artificial intelligence and human creativity. By giving AI the goals and constraints of your design, it can offer solutions and evolve concepts with the designer’s original intent. Autodesk and Van Winjen have collaborated on Generative Urban Design, where architects can provide basic parameters and conditions for AI to explore the possibilities for residential areas.
Design Insight (Advise you)
PTC Creo will automatically check the validity of 3d objects. If the surface of an object is incomplete, the software will advise and reject uploads to the cloud. CFD Direct allows designers to analyze the state of fluid mechanics in real-time while drawing 3d models.
Via CFD Direct
Simply put, the integration of AI is helpful for designers. As a contemporary designer, there are three things to stay ahead of AI: consider the new functions of software, learn the programming languages, and emphasize the value of reeducation. AI is smart, but can’t judge what is beautiful.