Architecture Beauty with AI
Author:
Celestine,- (s3524864)
Elvina Amelia (s3428572)
Qiannian Ye (s3687733)1. ABSTRACT
(TBW)
2. INTRODUCTION
We are in an age where we are increasingly dependent on artificial intelligence. (give a relatable example. I.e: mobile phone or internet data mining). Since 2006, the popularity of artificial intelligence is rising day by day. Even giant companies such as Google and Amazon think that artificial intelligence is the future. Thus, this amplifies the conception of artificial intelligence integration into most of the things we do in the days to come, without any human interference.
Besides its undeniable presence in our daily lives, there are also observations of an increasing popularity of artificial intelligence being integrated into design and architecture in recent years. However, it has been used mostly as a performance optimization tool, such as BIM software.
Yet, optimization is not the main focus of architectural design. Architects are responsible for more than building performance. Architecture has always been one of humanity’s cultural embodiment. This is mostly shown in the aesthetics of architecture in a certain period of time. Thus, the aesthetics in architectural design is a crucial topic to discuss during this changing time.
This paper is examining the beauty that is produced by artificial intelligence in architecture through the speculation that AI will be fully autonomous and instinctive.
3. ARTIFICIAL INTELLIGENCE (images to be added)
The research on artificial intelligence started in the 1950s and it is dictated by an algorithmic input. Between the 1950s till now, many technologies such as smartphones and computers were created and used by humans where most of them function through a script that allows them to work at its optimal. Hence, this optimisation creates an expectation/ impression to most people that machines works in perfection and errors only come from a human.
The idea of teaching the computer to think and behave like a human has been conceived since its invention. In the 1950s, the first study of machine learning using an artificial neuron network has been developed and in the late 80s, researchers were able to successfully overcome the limitations from that first attempt by creating a more complex multi-layered neural network.[1] At the time, it was able to recognize handwriting and translate it into digital data. Nowadays, AI owned by Google can translate a physical magazine written in a foreign language in real time by pointing your smart phone’s camera to the words on the pages.
Current breakthrough of artificial intelligence in producing usable outcomes are reached through machine learning process, where AI is fed a massive amount of data that researchers wish it will learn. Then the AI can compute the regularly repeated patterns from all the data it received and learn to recognize them. However, machine learning for AI is the easy part.
For it to be able to produce a satisfactory result, AI needs to go through a deep learning process, in which it practices what has been learned to create its own version. To learn like humans do, trial and error is the heart of this process for AI. The positive outcome of these steps is vital to the new stage of automation.
When the machine can independently make decisions, most of human involvements can be removed from the equation. Autonomous AI will revolutionize architecture design once it possesses its own instinct.
Thus, humans brought forth this expectation of no errors towards artificial intelligence. The expectations towards its accuracy in performing are even higher because AI is instinctive compared to previous technologies. However, this ideal towards ai has neglected a lot of factors that it might make mistakes due to the possibility that ai will be affected by other frequencies in natural circumstances.
This relates back to our research on nanobots in the early stages of research for this paper.
Nanobots are robots that are measured in nanometers. Currently, they are in the research and development phase where the research is mostly in the medical field. Current nanobots are still working in an isolated environment where the environment is assumed as utopian. This runs in parallel with the neglection of other factors in the human expectations towards AI. The diagrams below will further elaborate on this argument.
(Figure 1: current nanobot photo)
As shown in figure 1, it is shown that the artificial intelligence within the nanobot is able to plan the best and most efficient path for the nanobot to move through its surroundings for it to function at its best.
(Figure 2: Current artificial intelligence advancement)
(Figure 3: Prediction of the advancement of artificial intelligence in the future)
The current advancement of artificial intelligence is reflected in figure 2. Since the research on artificial intelligence has shown a tremendous increase in recent years, we are speculating that artificial intelligence will be more advance in vision, robotics, planning, scheduling and optimisation when it is more rooted and established in architecture in the future as shown in figure 3.
Through this research on nanobots, we speculated that its artificial intelligence has a huge potential in the architectural field in the future. The diagrams below outlines our speculation of how the artificial intelligence work in a natural environment.
(Figure 4: Programmed path)
(Figure 5: Artificial intelligence has the possibility to receive unintentional sensory input such as sound waves)
(Figure 6: speculation on the possibility that the programmed path will be affected by other frequencies in the natural)
(Figure 7: speculation on the likelihood that the outcome of the architecture will be affected)
The artificial intelligence that is speculated will be applied to different mediums in the architectural field. Therefore, the amount of fluctuation will vary depending on the size of the medium.
This raises the question of how mankind perceives beauty. Beauty is an extremely complex subject in the field of design as it is often said as being highly subjective. Nonetheless, it is probable that AI can be taught what mankind find beautiful. However, a big question still remains. How does mankind measure beauty?
4. HOW DO WE MEASURE BEAUTY?
Since ancient times, people used mathematics as the basis of architecture. Such as temples, palaces, and vaults. The beauty of mathematics in architecture might have come from an extended period of math-based architecture. Math and physics have become the foundation of architecture for a very long time. Perhaps the application of mathematics in architecture has been deeply rooted in the minds of mankind. Thus, our perception of good design is favoured towards mathematics.
As mentioned before, Artificial Intelligence works with input algorithm, which is based on mathematical principles and logic. Math enables human to have a measurable system for beauty. It is also easier for AI to determine a workable parameter. So, it is safe to say that designs generated by AI will have mathematical characteristics. Thus, the measuring unit of beauty generated by AI should also involve a mathematical point of view as human measures beauty with math too and it gives human and AI the same base criteria.
(Figure 8: This artwork created by Google Deep Dream is selling for serious money. Via World Economic Forum. [https://99designs.com.au/blog/design-history-movements/artificial-intelligence/])
4.1. BEAUTY IN MATHEMATICS
Intro: (to discuss the current generation of computational architecture design. Example of work by Alisa Andrasek, Roland Snooks, Greg Lynn, etc.)
- such as tessellation - tiling of a surface without gaps or overlaps, creating a complex pattern.
The tessellations are defined by the coding of mathematics where shapes are arranged in repetition without gaps or overlapping each other. There are a few forms of tessellation, they consist of the tessellation of regular geometric shapes, fractal tessellation and Voronoi tessellation. Regular tessellation work by the regular geometric shapes repeats themselves back-to-back seamlessly. Fractal tessellation is more complex than regular tessellation because it functions in a pattern where a shape in multiple scales fit together. The figures below elaborate how fractal tessellation can work in a single plane and multiple planes.
- bio-generative pattern, Fractal
- parametric design - Greg Lynn, Zaha, Schumacher - spline model (Carpo, Breaking the Curve)
- generative algorithmic agency - Roland Snooks
- Symmetrical ornaments in high-res - Michael Hansmeyer & Benjamin Dillenburger
1. Golden ratio
"phi" (Golden Ratio or Golden Section): ={1.618}{033}{988}{7}\ldotsĪ¦=1.6180339887…
The golden ratio has been applied to various works by artists and architects since ancient
times.
2. Strong centers.
When space is divided into multiple parts, each part will have a new center. When these parts form irregular subdivisions, it is possible that the center of each small part is not the same as the original center position. The brand new centers will create completely different patterns.
(Figure 9)
3. Boundaries.
Boundaries are related to geometry. Two points can be connected into a line, plus the third point can be connected into a triangle. The triangle can translate into many kinds of polygons. The final format is close to the circle.
(Figure 10)
4. Alternating repetition
A series pattern such as 1, 2, 3, 4, 5...n+1.
(Figure 11)
5. Positive space
On a plane, some graphics create a negative space, such as an acute angular in a triangle. If rotated a little or add a new angle, the negative space can become a positive space.
(Figure 12)
6. Symmetries
When an item has its own symmetrical axis, it can easily complete the whole form. One of the outcomes of the mathematical beauty is the mirror effect.
(Figure 13)
7. Deviation
A slight change in the value can cause deviations. The error can bring diversity to the whole, and also vary in density, structure, quantity, and strength.
(Figure 14)
8. Void
A void is indispensable in architecture. Whether it is private or congregate activities, the generation of different void creates more conditions for people‘s activities.
(Figure 15)
However, there have been studies on the fundamental of beauty from the point of view of our brain. Neuroaesthetic is an attempt at solving the very complex question of the definition of beauty. It is possible that AI can assist us to finally come to a definitive conclusion.
4.2. COMPUTATIONAL DESIGN PROJECT: FINDING BEAUTY IN PERLIN NOISE
- Intro: Desire to TRANSCEND POST-MODERNISM AND INTO POST-HUMANISM ERA
Contemporary computational designers are working to transcend post-modernism architecture.
- Architects as designers: do the same job as AI - pattern recognition within an algorithmic script (examples of work by Alisa Andrasek, Maj Plemenitas, Hansmeyer/Dillenburger etc)
Architects working with computation and algorithm are doing what AI are best known for. Pattern recognizing.
- formal quality desired by humankind can be taught to AI to someday recognize. (disturbance in the pattern, unique isolated formal quality)
Within the pattern generated by an algorithm, architects are searching for qualities that can provide unique spatial experience for human, on the human scale and our point of view being considered. We are searching with mankind benefits and criteria in mind.
(Figure 16: Pattern recognizing exercise creating bigger voids for human habitation potential. [Elvina’s studio work, 2018])
- Currently: human-desired qualities from the algorithmic pattern are sourced from human spatial needs (human as the main agency). What happens when the main agency shift from human to inhuman?
However, in the future, we speculated that AI will have its own intuition and its own agency. As the main builder, it will eventually create space for themselves that might only able to be perceived by AI/machine.
- Conclusion: inhuman population might take over space and create a new aesthetic that can only be perceived by AI. (thinking process caused by design studio exercise)
- mankind has so much item in their possession or produces waste that eats up space on earth. Non-human matter already has agency in a way.
- we build for our use.
- what happens when AI build for us and themselves? It possibly has a very broad reach of scale and computing ability.
- AI will build space for them (conclusion), has its own agenda.
- AI perception/point of view & influences will be different from that of human’s so it will perceive beauty differently as well.
- The production of a machine exponentially increase
- Thus, the machine will have their own agency in architecture design and effectively shift architecture typology to provide space for them too and create new beauty in their perception.
It is true that AI can still have our best interest as its priority. However, we cannot deny that as it becomes more and more intelligent and possesses a different point of view than human, it will start to create a new aesthetic that mankind might not be able to grasp.
5. POSSIBLE OUTLOOK INTO THE FUTURE (AI-BEAUTY)
Presently, the existing artificial intelligence is mostly utilized as a performance optimization tool. However, this should not be the case in the field of design.
The vision of creating artificial intelligence was to make it fully autonomous and instinctive by itself. The foreseen characteristics of artificial intelligence allow it to remember things much more clearly thus allowing it to target a similar issue quicker than a human being. There is a high possibility that the nature of artificial intelligence can construct ornaments in a higher resolution because it is able to reduce the number of human errors compared to a human that is on the task. Artificial intelligence has a strong ability to learn and analyze the situation. For work, AI can accurately locate the route, choose the working mode rationally. This character that it carries also enables it to function in a greater precision when it is doing its designated job as it will not be distracted by psychological factors that humans have biologically.
Human has an expectation that AI will produce a flawless work, the work that is exactly the same as the one they imagined in their mind. However, it is not possible for the artificial intelligence to work in its programmed path under the site environment due to the other existing frequencies that are around. The fully autonomous artificial intelligence will create unexpected results due to the possibility of being distracted by other frequencies on site. Although the outcome is not what is expected by humankind, this plays as a role to define the beauty that is created by artificial intelligence.
6. CONCLUSION
We are going into an age where we are increasingly placing our faith in artificial intelligence rather than our own hands. The intention of creating AI is for it to possess the ability to make instinctual-based decisions like human beings do. There is a possibility that they will make their own decisions to create outcomes that are far different from the initial input. The possible instinct that they will eventually have and the impossibility of total perfection in a real-world application caused by other environmental input will thus create beauty that humans are unable to comprehend.
SOURCES / READINGS
[1] “The Rise of AI”. Bloomberg. https://www.youtube.com/watch?v=Dk7h22mRYHQ
http://www.arnopronk.com/bestanden/Bridges.pdf
https://www.artforum.com/print/201402/breaking-the-curve-big-data-and-design-45013
Picon, A. (2013). Ornament: The politics of architecture and subjectivity. Chichester, West Sussex: John Wiley and Sons.
Carpo, M. (2011). The alphabet and the algorithm. Cambridge, Mass.: MIT Press.
https://99designs.com.au/blog/design-history-movements/artificial-intelligence/
No comments:
Post a Comment