We are at an inflection point in a technological revolution that has the potential to change the way we do everything, and it's important to pay attention to relevant developments. In this article, the author shares three roles of service designers in the AIGC era, let's take a look together.
We are at an inflection point in a technological revolution that has the potential to change the way we do everything. Generative AI will help us co-create new products, services, and experiences. In the development of these new systems, service designers can play at least three roles.
First, it will shape the way people understand and interact with AI. Next, service designers can bring their "lenses" to help ensure that these systems are useful, usable, efficient, effective, desirable, and differentiated – from both individual and provider perspectives. Finally, our best service design methodology and foresight can be applied to responsible design, avoiding bias and unintended consequences.
Generative AI-Generative AI
A generative AI system is a type of artificial intelligence that generates new content based on the probability of relationships or data learned from large dataset examples. These systems are often used for tasks such as generating text, images, audio, video, and even music. They work by using machine learning algorithms to learn patterns in training data and then using that knowledge to produce the output of the original structure, not just a copy of the input data.
As a user, we can enter a few words and the response returned can be a paragraph, a detailed illustration, a video, audio, and much more. It's a fascinating relationship – we give little, but we give a lot. This is one of the reasons they have such potential, but can really demand our talents as service designers because they become partners in our collaboration and co-creation.
We are in this moment of transformation as the AI community has developed large neural networks that can "train" millions and billions of web-based text fragments, images, audio tracks, and more, all thanks to continued exponential growth in computing power and networks. The resulting trained system is unimaginably fast and benefits from an extensive dataset of human knowledge that is constantly uploaded to the network.
Both ChatGPT and DALLE are service systems from AI research and deployment company OpenAI. ChatGPT is a leading example of a text-to-text generation AI program. DALLE IS A TEXT-TO-IMAGE GENERATION AI PROGRAM. Texttoimage programs can also be used to generate anything that can be represented as an image, including sound and music and even chemicals.
We are already seeing useful applications in the life sciences, and one promising area of research is the production of proteins to meet specific constraints in drug development research.
The applications seem endless. They have the potential to support all forms of cognitive work, such as doctors diagnosing diseases, prescribing treatments, and, more mundanely, managing their own workload. It can be anything from personalized copywriting to qualitative research analytics to cybersecurity. Perhaps the most advanced application of text-to-text systems today is GitHub's CoPilot, a program built on ChatGPT that helps engineers write software code.
CoPilot can also help engineers generate comments that describe the functionality of existing code. Assistants or so-called "intelligent agents," such as CoPilot, can quickly appear to do a variety of jobs, from managing buildings to flying airplanes and more. Reid Hoffman describes this support as having a "co-pilot" responsible for everything that expands the relevance of human knowledge and activity to the TikTok level.
We will soon reach a point where technology will get us to the level of accuracy and speed of delivery we need to revolutionize the work we now know and potentially democratize AI for all of us to use on a daily basis. Mira Murati, chief technology officer at OpenAI, recently said, "We're trying to build these generic systems that can think about the world in a similar way to humans; So the system has a powerful concept of the world."
Can these systems generate content based on patterns and relationships in a way that is not much different from how humans design today? As we begin to explore the application of generative AI in designing new product and service systems, it's important to consider the benefits we gain from this technology and what we may lose in the process.
Generative AI systems have the potential to help service designers radicalize from the start. The unexpected serendipity that comes with experimenting with an AI-driven sparring partner can lead to higher-quality results while helping human and machine teams collaborate better. They will help service designers challenge and refactor existing practices. But these new systems could also lead to a darker future.
In this new collaborative environment, how will we determine that we have designed the essence that is important to solving human needs? When everyone can participate in creating a high-fidelity prototype of a concept, how will people judge the outcome?
Today, these systems are mostly shaped by technologists trying to move them forward as quickly as possible, but the work is still in its infancy. As designers, while we are not directly responsible for the implementation of the AI model itself (at least in the short term), our involvement is critical in shaping the environment in which we choose to generate AI as a tool, experience, and peace. Use – Used to design the final product or support people in other work.
It's time for service designers to step in to make sure we're meeting people's needs and have a clear understanding of what it means when technology acts on behalf of users – and all designers – in these applications.
Service designers and AI
Service designers play at least three roles in shaping the development and use of generative AI as a new technology:
Role 1: Design new ways to interact and understand AI
The first role has officially begun: we can help shape the development of generative AI by using our design research methodology and expertise to understand what people want to get out of their experiences and how they want to use the technology. We can then work with technical experts to create systems that can deliver those results.
As designers know well, there's no point in creating an amazing product or service if it doesn't meet people's needs, or if they can't effectively accept and use it in their daily lives.
Designers excel at showing "what we think" when presenting scenes, future journeys, or any type of sketch or rough prototype. We match the fidelity of the prototype to our audience and stage during the design process; While we were still forming ideas, it was rough at first, and as we narrowed down our design decisions, more were done.
Mike Kuniavsky calls these outputs "partial representations of potential futures designed to give us a glimpse of the meaning of our ideas." Kuniavsky continues, "When they [AI systems] do the details, they shorten the value that using local abstractions adds to the creative process. Filling half-baked ideas with partial, probable, but unreal results is exactly what people do when they create, just as they imagine. ”
We design with just the right amount of detail to encourage our colleagues, clients or their customers to contribute together.
Having a conversation when collaborating between people or AI systems on partial representations is fundamental for teams to create shared context. Adopting our participatory approach ensures that systems are co-created and co-produced.
This isn't just an inflection point for AI – we need to redefine the user experience. Providing the right apps and interfaces, making the experience engaging and simple, and feeling the "magic" of change without over-provoking, will be the minimum we require. ChatGPT is a conversational tool that enhances people's ability to explore options. There will be countless others.
Our participatory approach can also help ensure that these new applications are skillful, not deskilled. When designers are given multiple design options suggested by AI, their craft shifts from "pure" creation to curation on top of curation, to final creation or version. These faster iteration loops have the potential to create a virtuous cycle of learning for service designers and end users.
How much technology do we need to know to believe and harness this incredible ability to simplify our lives?
Imagine a doctor asking why an AI system suggests a specific treatment path. We can help physicians build a robust model, relationship, and data model to support the situation they've just encountered. Designers are good at zooming in to see details, and are also good at explaining the whole from a distance. We can use this skill to help design "interpretability" so doctors can scrutinize logic and results, including visualization and fact-checking of all references.
As a subfield of artificial intelligence, the concept of explainable AI has been around for a long time, focusing on developing and improving the interpretability, transparency, and comprehensibility of AI systems. But there's still a lot of work to be done to define interpretability in these new build environments. The goal is for humans to understand how and why AI systems make certain decisions or produce certain outputs.
The new focus will shift to asking questions such as "Why does the digester emphasize that part of my paper?" " and so on.
Making AI easy to understand can help increase trust in AI systems, reduce the risk of bias (depending on the training data used), and improve the overall performance of AI systems. Explainability will also give people an understanding of the strengths and limitations of the system and how we can best use them in different environments.
Role 2: Bring a designer's perspective to business growth and differentiation
More than thirty years ago, C.K. Prahalad and Gary Hamel wrote an influential Harvard Business Review article, "The Core Competitiveness of the Company." Core competencies are the resources, skills, and learning that make an organization stand out in the marketplace. These new AI systems will force businesses to look at their resources, skills, and learning through a new lens of business reinvention.
For a long time, the goal of service designers has been to produce service interfaces that are useful, usable, efficient, effective, satisfying, and distinctive for individuals and providers. Fundamentally, they need to understand which common applications don't become differentiators for businesses and which are at the core of their business capabilities, so there is potential for real differentiation where designers can help.
In addition to shaping interactive experiences or explaining the reasons behind AI making recommendations, we can work with clients to design how businesses respond to change. Through design research, we can gain insight into what matters most to people and connect what we learn to the purpose and core competencies of the organization. Not only can we help determine where and why accumulated data brings value to AI models, but we can also seek to improve relationships between the people involved.
Businesses will need to focus and decide what data and models to have and the digital twin to build so that the two compound over time, grow together, and create competitive advantage by expanding the company's goals and key differentiators.
Of course, banks will want to collect and keep data about underwriting. They can devise a way to explain instant loan decisions, but what will fundamentally enhance their relationship with their customers? People have different mindsets when it comes to money, and these mindsets change depending on the circumstances.
Can apps like ChatGPT be used to customize interactions to create personalized advisors to nudge when you want to remind you that you don't want to go over budget, but sensitive enough to cheer you on special occasions? Shopping activities'? Each interaction helps form a learning loop and continuously improves their AI models while combining their differentiation, customer relationships, and the bank's bottom line.
It is not enough to rely on past capabilities. For example, if your business is already using machine learning to drive your decision-making, you'll need to change the way you do things to integrate these more advanced systems. You need to design your product or service to grow, take your industry to higher and higher levels, and seamlessly meet the untapped needs of your customers.
We are currently developing ways to help our customers seize these opportunities.
To act quickly, you need to be able to assemble a high-performing team of data scientists, AI experts, psychologists, learning specialists, business strategists, designers, and design researchers who share a common approach and language.
Internally, introducing generative AI-powered design in your organization will require open change among multiple stakeholders. If it can propel your business forward, it will most likely completely disrupt the traditional structure. Service designers can take the lead in promoting this heightened focus on differentiation and shape goals with these diverse teams.
In order to facilitate and mediate collaboration, a new perspective on the system is needed. Connor Upton sees it as a mashup of service blueprints and system architectures. A view that predicates organizational IP value, governance value, and customer value.
3. Role 3: Design governance, safety, and ethics by co-creating with technicians
These teams will have a significant responsibility to design right from the start. They need life-centered design principles and apply them when conceiving systems. This means co-creating as a multidisciplinary team from the start to design for the desired outcome, while leveraging foresight to continually understand the unintended consequences. As Mira Murati suggests, its goal is to "iteratively build mitigations."
Academic institutions are already working in this direction and are calling for the establishment of a new version of the Institutional Review Board (IRB). At Stanford, it's called the "ESR," or Ethics and Social Review Board.
As designers, we like to say that we represent people and contributing to these efforts can make this a reality, but as FeiFei Li, co-director of the HumanCentered AI Institute at Stanford University, says, "Safety is one of those words like health: everyone wants it, but it's really hard to define it." ”
Quickly establishing ways to ensure fairness and diversity in AI-generated products and services requires a lot of consideration and work. In addition to interpretability, ensuring that the data set at the heart of the system is fair will also be key.
What input parameter sets or training datasets were used? As datasets grow exponentially, year after year, can we use blockchain to determine where they come from and know what's "raw" and what's generated? Without checks and balances in place, says Melissa Heikkilä11 of MIT Technology Review, "we may be witnessing the birth of a snowball of nonsense in real time." ”
There are also some major considerations for how AI will affect people and the size of their jobs. Can we leverage and address specific labor shortages and challenges at scale on a global scale? Or consider using AI capabilities on a small scale to help children learn by iterating on input to AI so that the result is an interactive experience rather than a "final" system output?
Can we teach our children a subtle appreciation for co-creation (human or AI) and develop a different appreciation for their unique endeavors? So far, there are many problems and few clear guidance. Scenario building is at the heart of service design, and we can use our forward-looking approach to look for early signals and identify trends in real time to explore multiple possible futures and mitigate them iteratively, as suggested by Mira Murati.
Service designers can represent people – as part of an interdisciplinary team that applies principled design from the outset to address the ethical, governance, and safety issues of AI and designs for the desired outcome, but this will be difficult.
Generative AI will revolutionize the role of service design in both the public and private sectors
To sum up, there is no doubt that as a tool for generative design thinking, generative AI can help service designers be bolder and more efficient in their work. We need to adapt and develop new skills and expertise in areas such as machine learning and data science to take full advantage. At the same time, the use of generative AI in design has the potential to change the role of service designers.
We can work with technical experts and these broader teams to streamline people's experiences – in the way we've always done in our designs – suggest ways for AI to remove steps, and come up with nextbestmove suggestions to make it a simple and stunning aesthetic experience, and so on.
In this revolution, we can become design optimists, collaborating across disciplines with those who shape the tools to shape the future we want. Marshall McLuhan once said, "We shape our tools, our tools shape us." We now have a once-in-a-lifetime opportunity to shape these tools, and if we do, design and service designers will do better.
Shelley Evenson (this article has been licensed by the parties and the original author)
原文名称：Generative AI Needs Designers ｜Three roles for service designers
Translator: Chen YuzhiYeutz Chen, WeChat public account: YeutzDesign (ID: Yeutzsheji), focusing on the field of service design, committed to the research of service design innovation and transformation.
This article was translated by @陈昱志Yeutz Chen and published on Everyone is a Product Manager. Reproduction without permission is prohibited.
The title image is from Unsplash and is based on the CC0 protocol.
The views of this article only represent the author himself, everyone is a product manager, the platform only provides information storage space services.