To the limit, and beyond within AI’s Impacts on Creative Design
Every revolution is not radical. The paradigm shift has always permeated to our daily lives. At the point, when these little shifts hit the singularity, humanity has always adapted themselves to ‘new’, and the majority gets to live the ‘new’ too.
In this regard, ‘Artificial Intelligence’ can also be the ‘new’. The biography of ‘Artificial Intelligence’ is indeed short. It has begun from a trivial computational machine that supports complex and vast quantities of numerical values. But now it became a machine that can compute complexity way much faster than any mankind can.
Constant research in academia enabled the machines to perform in design, which requires more complexities in the calculation. This consistent effort in the development of AI has brought the commercialization of its function, and it is broadly used in design industries.
AI in the design Industries
For instance, Spacewalk, the South Korean Real Estate Protech startup developed an AI called ‘Landbook’. The machine constantly learns geospatial information and land laws through ‘deep reinforcement learning’.
As long as user input
▶Maximum area
▶Maximum expected rental income
▶Maximum lighting rate
to the machine, it gets to plan the most suitable design for the site selected by the user. Moreover; the machine also gets to calculate ideal numbers of tenancy, expected construction costs, and expected profit from tenants.
Such practice has allowed automation in the architecture industry, which helps the cooperations to think of the most efficient way to make a profit. However, this is also being criticized at the point that the architecture is losing its meaning in terms of design.
However, the design planned by the ‘Landbook’ has meaning to its design. The meaning has considered all the economic parts in the architecture, which is also an aspect regarded the most by active architects. In product design and architecture fields, aesthetical value is evaluated by how have problems are solved through design. In this sense, the machine had almost perfectly satisfied what is expected in the industry. In fact, it has been widely used both in the public and private sectors.
This can be inferred that the AI-assisted process can actually help society to enjoy benefits out of it. At least, the fact that AI-produced design work is getting recognized by architects, and it has been proven by the practice as I mentioned previously.
Using training model as a beginner
Speaking of the design machine, my group developed a GAN model, supported by Runway ML, to inspire clothe designers by generating its fake image. It was called ‘ThisFashionDoesNotExist’. The way how the GAN model works resembled how ‘Landbook’ works.
At our first trial, we failed to generate the fake clothe image. Due to technical issues in image resolution, it was created odd creatures that are not even familiar with the normal wears.
At the second trial, we were able to develop a model that can generate images in normal shape. However, the images generated were not distinct enough.
Although the first trial seemed to fail to generate a recognizable creature, compared to the generation from the second trial, at least it was able to show that the AI can create new value as long as we input more precise data to the machine.
This statement can be supported by the very first product we made as in a group. Which is the ‘Get The Right Information’, using the feature-recognition technique.
With existing knowledge and techniques, we were able to create new value for the users. Although the machine did not learn the dataset given, we were able to seek the logic and its potential to generate the ‘new’, which are also seemed to be found with the AI. The creative process here, even though it lacks creativity in its result, also showed a potential that the AI’s products may also be recognized as artwork to a certain extent.
Taking all these cases into the counts, I am certain that works generated by machine learning will create new values. In terms of speed, it has been a while that the computation excelled in it. Yet, the products produced at the moment are not going to be valued as art pieces. On the other hand, when time flies, the creations made by machine learning may bring a paradigm shift to society, creating new values as it has been. One of the fundamental principles in design is to solve the issue through design, and AI has achieved that to a certain extent at the moment.
Therefore, even if the machine excels at generating value faster than mankind does, the results will be treated differently. Meaning that we are going to co-exist with the machine, while their products are being respected as they are. Like how Gogh’s works were judged badly back in his era, but they are recognized as one of the best artworks in human history. So yes, this is definitely the ‘new’ we are going to deal with. The products that are produced by AI can be counted as design work, but need more time for the machine and public to get used to the ‘new’.
