This article has a total of 3198 words and is expected to read in 8 minutes
TCC Intelligence Bureau's Title 196 Dry Goods Sharing
Part 25 of 2023
TCC recommendation: Hello everyone, this is the TCC Translation Intelligence Bureau, and I am Zhang Yutong. As a professional with rich experience in AI product design, the author introduces 7 principles she summarized from her experience in designing AI products for us, which are worth considering carefully when designing AI products.
In today's design community, there is too much discussion around what artificial intelligence (AI) entails. How many possibilities does AI bring, and how to use it, incorporate it into your own design process. At this point, you might be asking yourself, "What does it mean to work with AI as a designer?" How do I design for AI? "You're not alone in having these confusions.
When I started working on AI products a few years ago, I had a lot of confusion, which is why I've dedicated myself over the past few years to sharing some of the key principles that have led me and my team. Once you master these skills, you can begin to develop the experimental mindset you need to use in the teams working on AI products and accelerate the value of your users.
Here are 7 principles you should keep in mind when designing products with AI:
1. Start with a clear goal
1. Start with Clear Goals
When designing a product, you can't rely on AI or machine learning to find the problem that needs to be solved. For your solution, it's important to start with clear and specific goals that align with business intent user intent. This will give your team and product a strong sense of purpose and direction.
For example, in 2018, Google Docs cited machine learning models to suggest fixes for grammatical errors in real time as users type. When the model was very accurate at checking for grammatical errors, users quickly realized that the model's suggestions did not fit their writing style or way of speaking. After spending millions of dollars, Google realized that the accuracy of the grammar provided by the model was not as important in the minds of users as it was to provide suggestions that the user was trying to convey consistently. In this case, the logic is simple: AI can provide powerful ways to solve complex problems, but the ultimate success of AI depends on how it aligns the specific goals and needs of your products and users. That's why you need to define your product goals based on the user's perspective and ask: How can AI help us solve this problem?
2. Form your algorithm based on your design decisions, not your design decisions based on algorithms
2. Shape Your Algorithm with Design Decisions, Not the Other Way Around
When designing AI products, you must consider diverse design decisions to shape your algorithms. For example, Facebook's What's New feature.
What's new at Facebook uses machine learning algorithms to show users content that is more relevant to them. However, someone needs to decide that the news stream is scrollable, that people should be able to respond to posts with emojis, and that the content needs to be presented in a specific style. These are examples of how product and design decisions do not rely on machine learning. In other words, when what's new is a product that uses machine learning, there are decisions that contribute as much to the success of what's new as machine learning techniques. By thinking carefully about these decisions, you can form your algorithm to meet the needs of your product, rather than forcing your product to adapt to your algorithm. This way of using decisions to form algorithms makes the product more efficient, effective, and user-friendly.
3. Provide a channel for continuous feedback
3. Create Avenues for Continuous Feedback
Designing with AI requires a constant feedback loop. You need to get feedback from users and incorporate it into your product. This helps you improve your product and provide value to its users. To build a feedback loop, you can design an experience that serves as a reward for your user education algorithm.
Facebook's What's New feature is another great example. What's new allows users to generate feedback on posts, which can give clear signals that users want to see content, and can be used to further refine interest-based news feeds. You can also use a variety of feedback mechanisms, such as questionnaires, user tests, and some analysis tools. By prioritizing user feedback, you can create products that respond more quickly to the evolving needs of your users.
4. Adaptive design
4. Design for Adaptation
It's important to consider how your product interface should accommodate user feedback and provide value in any user scenario. In some cases, any new piece of information provided by a new user should affect how quickly your AI product can present the latest results. As an example, consider Google Maps' location-based reminder notifications.
Image credit: Tech Crunch
any action you make; For example, going to a new neighborhood or searching along the route will give you new information. That's why you should understand the necessity of "speed" for your product. By adaptive design, you can provide a more resilient, flexible, and enthusiastic product for each changing need of your users. In AI terminology, this can be done through many technical means, such as machine learning, dynamic algorithms, environment recognition, scenario analysis, and customized recommendations.
5. Embrace improvement and human help
5. Embrace Refinement and Human Help
While AI can automate many decision-making processes, it is no substitute for human judgment and expertise. When designing user flows for your AI product, it's important to consider how humans will participate in the AI decision-making process. You should leave a door for users to improve the output of the AI and provide human help when the user needs it.
For example, Amazon.com allows users to remove products that may negatively impact recommended products. Imagine you just bought a toilet ring on Amazon and don't need to buy another one in a short period of time, won't you get bored with seeing toilet ring deals in your future recommendations all the time? Because of this, you should prioritize "design for opportunities for correction from users" features. You can further this through a variety of technologies, such as AI training systems, decision support tools, and expert systems.
6. Contextualized design
6. Design for Contextualization
When designing AI, context matters. You need to consider the context in which your product will be used and make sure that your AI product and users remain relevant in that context. For example, in the search experience, the best recommendations are not enough.
Think about it, the same word may mean different things in different contexts. Search engines like Google use machine learning to tailor your search results based on your location, search history, and time of day. So ideally, the results provided by your algorithm need to be always contextualized to help the user. And your mission as a designer is to reinforce this through more contextual perception of the user interface. Designed for context and relevance, you can create products that provide value to your users in any scenario.
7. Learn more in my next lesson
7. Learn More in My Next Class
You're excited to continue learning AI x product design, aren't you? Join me for my next lesson and pick up where we left off. I'll share some more practical tips that have led to successful innovative products for me and my team. So come and say hello and let's enter the world of AI design together!
Ultimately, designing for AI is about putting people first. It doesn't necessarily just need to be related to data, but it needs to be related to the data insights provided to your team. So think of data as the engine that starts the plane, and insights as what travelers really care about. By adhering to these 7 principles, you'll create AI products that not only innovate, but truly deliver value to ordinary people.
Original: https://uxplanet.org/7-ways-to-get-started-designing-for-ai-ml-products-bef764ca1b27
Written by Lola Salehu
Translator: Chen Yuzi
Reviewed: Li Zehui
Editor: Han Shuo
The translation of this article has been duly authorized by the author (screenshot of the authorization is below)
Featured articles from previous issues:
Ten tips to help you design an e-learning app
This knowledge of color psychology teaches you how to convey information
Case Study | Cornell University Side Hustle Community Website Design
Web Vitals – Google's next-generation web performance experience and quality metrics
What can web designers learn from Japanese aesthetics?
Case Study|A food app that brings you an unforgettable experience
Here's a complete instruction manual on how to create a portfolio that impresses an interviewer
How to do a good UX project? Start with a good plan
How to set up a design system
How to grasp the needs of users at different levels: return to the essence and polish the information architecture
TCC Vision|2021 UX Design Trend Analysis
- Designer Self-study Community -TCC Design Intelligence Bureau welcomes friends to join us to exchange design knowledge, understand global design information, exercise English ability, and explore more possibilities~
Add a small assistant WeChat and note "community" to join the readership.