anticipate

Artificial Intelligence
Design Futuring
Human Reasoning
Reflection

Design and AI

This thesis sees design as a way to make a positive socio-economic change. It investigates the influences of artificial intelligence on design processes, trying to understand how AI can augment instead of making choices for us.
see more
minimize
"Design has too often been deployed at the low-value end of the product spectrum, putting the lipstick on the pig." - Dan Hill.

In addition to the quote above, Dan Hill adds in his book: "In doing this, design has failed to make a case for its core value, which is addressing genuinely meaningful, genuinely knotty problems by convincingly articulating and delivering alternative ways of being." In the spirit of the quote, this thesis will address the part of the design that deals with intertwined but meaningful problems.

Herbert Simon framed design as developing courses of action to transition from existing circumstances to preferred ones. Simon's definition encapsulates the essence of design as an agent of change. This definition sets design as a discipline beyond simply addressing aesthetics or functionality but to actively shaping the environment and society.

This research explores the diverse influences of Artificial Intelligence on the design process, particularly the ambiguity and complexity characterizing the initial stages.

Navigating the Fuzzy Front End

Design argumentation is always guided by the critical question of what information is important and what might have been overlooked. anticipate aims to navigate this uncertainty and improve decision-making by highlighting potential missing knowledge and biases.
see more
minimize
The process of design argumentation is primarily guided by determining the critical information and identifying any overlooked data. The goal is to navigate this uncertain terrain and enhance decision-making by underlining potential unknowns and biases.

Designers consistently encounter a multifaceted environment that challenges them to use more than just problem-solving and analytical skills. Creating effective solutions requires the identification, framing, and iterative redefinition of challenges. The complexity usually stems from the uncertainty about the significance of particular information and the inability to detect unknown factors timely.

Research consisting of 14 initial interviews was conducted under the premise that AI could significantly simplify the design process by reducing its complexity. The interviews revealed that there are well-established methods for summarizing and clarifying information, primarily through divergent and convergent strategies.

Complexity emerges when there is insufficient knowledge about the subject. The initial phase involves extensive questioning among stakeholders to eliminate these unknowns. Gathering a primary set of ideas, hypotheses, and potential issues significantly addresses this challenge.

Designers frequently find it challenging to dissect problems, mainly when time constraints restrict research. Users may sometimes adapt to specific issues, compounding the difficulty in identifying problems. A systemic view of problems is optimal, but the inherent complexity often makes this approach impractical.

The process of decision-making is a crucial aspect that requires significant attention. Making decisions iteratively and transparently is critical. For this purpose, designers can effectively condense issues and create clear presentations for stakeholders.

Locked in the Present

When used in the design process, AI consistently delivers the most probable outcomes, using its ability to analyze and predict on vast amounts of data. Using human-created data, the large language models are based on current thought structures and thus project the current beliefs into the future.
see more
minimize
There are inherent problems in using AI in design. Biased human data could lead to undesirable results, which makes the propensity to prioritize the most probable outcomes detrimental to design.

The large language models are founded on data that tends to retain our societal thought patterns, implying that AI alone cannot solve problems that need a shift in thought structures. Complicating the matter is the existence of self-reinforcing loops, further solidifying societal thought patterns and influencing actions and expectations that drive our communities.

The training data of AI represent human perceptions and interpretations, as explained in "Preferable Futures" by Nohr and Kaldrack. They note that "data-driven AI exposes the prejudices and wishful thinking of those who feed it, thus stabilizing social structures and expectations. AI has been used in various contexts to limit uncertainties, whether in decision-making systems, training simulations, or full enterprise simulations. These applications share a common goal: controlling or making contingency controllable, leading to a rationality of "predictability.""

In business and economics, simulations designed to reduce uncertainties already impact our society's operations. They go beyond just scientifically predicting the future, instead suggesting, directing, manipulating, and creating futures premised on a belief in continuity, thereby stabilizing trajectories and path dependencies.

Preferable Futures

However, societies aspire to advance and aim for more preferable futures. Believing that humanity should still be allowed to decide on desirable futures, artificial intelligence should not influence our decision-making processes.
see more
minimize
The use of AI and large language models in design processes should be mindful of society's aspirations for progress and innovation. Existing beliefs, whether embodied in AI or in humans, should not impede our ability to envision and work towards preferable futures. This requires a conscious shift away from relying solely on the most probable outcomes projected by AI.

The demand for a more comprehensive societal transformation, as necessitated by crises like climate change and species extinction, requires rethinking these profoundly impacting technologies. AI struggles to address challenges that require a paradigm shift, mainly because it operates within and potentially accelerates the established thought structures of our society.

To summarize these complex matters more simply, there is a quote by Rolf F. Nohr: "Or to put it more bluntly: the (uncertain) future imploded into a kind of "feedback-effected present" in which tendencies are intensified or subdued. The future was hedged and immobilized". The fundamental idea is that we have no absolute independence of action since our decisions are based on the present thought structures.

Beyond the Surface

But developments into better futures are also slowed down by artificial intelligence. Humans also maintain current trends because of their existing beliefs and assumptions.
see more
minimize
Our perception of the world significantly influences the design process and outcomes. As an integral part of the design process, assumptions and biases can influence the course and outcome of the design. Cognitive biases, defined as systematic patterns of deviation from rational judgment, can affect decision-making, problem-solving, and, subsequently, innovation outcomes.

Mental models are cognitive frameworks that provide a simplified representation of how the world operates. These frameworks underlie our beliefs and perceptions, shaping our behavior and influencing our approach toward problem-solving and decision-making. Derived not from facts but from individual or collective beliefs, mental models embody what humans know, or perceive they know, about a system. The challenge with mental models arises from what they do not account for, making certain aspects of our world invisible.

Developing an awareness of these limitations and examining them critically is increasingly essential for innovative and effective design. This perspective aligns with a quote from Albert Einstein referenced in "Preferable Futures": "We cannot solve problems with the same thinking that we created them with." The quote encapsulates the essence of this thesis - the need to challenge assumptions and biases in design thinking processes and innovate beyond them.

Self-Perpetuating Design Decisions

Our decision-making is fundamentally based on our beliefs about our environment. These assumptions are reflected in our designs, which in turn impact our environment, reinforcing our initial beliefs.
see more
minimize
The self-reinforcing nature of design decisions, mainly how they are shaped by societal thought structures, can be very dangerous. Understanding these cycles is critical, especially when addressing systemic challenges such as climate change or social inequality. This necessitates reviewing the underlying thought processes guiding design decisions for innovation.

It can be beneficial for designers to be critically reflective and recognize how their choices contribute to these feedback loops. Additionally, they may understand the broader context and implications of their decisions. This involves questioning the assumptions, norms, values, and perceptions that underpin thought structures. Design professionals must embrace a more holistic mindset, considering the immediate implications of design choices and how these decisions will resonate across time and societies.

Example

For example, if we think cars are the best means of transport, we design our environment with more roads. This, in turn, makes cars a more effective way to travel, as there is more infrastructure to support them.

Idea

anticipate aims to challenge people's ideas critically. This helps uncover unknown factors in the design process and identify implicit assumptions in the design argument.
see more
minimize
AI has the potential to enhance idea generation in design by critically questioning the thought structures behind concrete ideas without influencing the ideas themself. Its extensive and differentiated data can help draw the designer's attention to new aspects of the process.

By challenging current thought patterns, we can break cycles and create space for preferable changes not constrained by existing thought structures.

The intention behind this is to identify gaps in the design argument that can arise due to the mental models, assumptions, and biases of the designers. More specifically, the application should support the discovery of 'unknown unknowns,' which the designer inherently cannot recognize alone.

Augmenting by Reflection

As a result, language models do not direct our decision-making to the most likely outcomes. Instead, they challenge existing ways of thinking and create room for innovation.
see more
minimize
The responsible use of AI should be ensured by not generating decisions or ideas but by providing stimuli for reflection that go beyond what the designer could consider unaugmented. This augmentation allows for a broadening perspective, enabling designers to challenge their inherent biases and assumptions. Reflection and questioning interrupt the manifested self-reinforcing feedback loops by exposing and questioning underlying assumptions before decision-making.

The focus is to reduce the complexity of the design process, not by filtering out valuable information but by enabling designers to manage uncertainty better, enhancing their capacity to transition from existing situations to preferred ones.

Certainty

Using AI in this way aims to lessen the uncertainty in the early stages of design and promotes a more complete view in the design process.

Purpose

anticipate aims to reveal potentially harmful hidden assumptions in design, promoting more holistic thinking and guiding us toward more preferable futures. This thesis explores this potential and serves as a basis for further discussions on the responsible use of AI in design processes.

Prototyping

By coding a prototype, the hypothesis of the research was made more tangible. It could thus serve as a basis for further interviews. In addition to potential dangers, the applicability, potential, and uncovering of Unknown Unknowns were found in user tests.
see more
minimize
Users could input their ideas and evaluate whether the results were relevant to their thought processes. This immediate interaction and feedback were beneficial in understanding and adapting to user needs and expectations. There was also a distinct shift in the discourse regarding the content of the thesis. The applicability of the results became more verifiable, allowing for a more in-depth and meaningful discussion.

Objectives and Constraints

The user initially inputs the objective and the constraints in the upper left corner of the interface. This process helps the user comprehend their goals while simultaneously providing the language model with more context.

Presumptions

After inputting the initial idea, the language model can generate relevant presumptions. These get broken down from a vague statement into tangible, verifiable conditions that can be answered through interviews or prototypes. The structure establishes a clear map with branches that organize the necessary validations.

Transparency and Adjustability

The right panel is designed to educate users on how the language model operates. This is intended to improve the transparency of the model's output and align their expectations with the outcomes. Furthermore, the users can define the preferability of futures by, for example, using the Sustainable Development Goals to challenge their ideas.

Collaboration

The interface provides a function that lets users integrate their personal discoveries and assumptions into the tree. Including a collaboration feature enriches this process by enabling teams to reveal and discuss individual beliefs and assumptions. This could lead to a significant improvement in aligning stakeholders.

Highlighting

While navigating, we might come across critical presumptions. These can be highlighted by us to emphasize their significance for the project. By doing this, we also convey our priorities to the language model, increasing the probability of receiving relevant results.

Custom Views

For a more comprehensive understanding, it is possible to create additional views for the tree, where presumptions can be prioritized based on personal criteria.