Dan Saffer
Designer, Author and Assistant Professor at CMU HCIIDan Saffer is a designer and the author of four books: Designing Devices (2011), Designing Gestural Interfaces (2008), Designing for Interaction (2006, 2009) and the best-selling Microinteractions (2013).
Since 1995, he's designed devices, apps, websites, wearables, appliances, automotive interiors, services, social networks, and robots. He’s worked at and for such companies as Twitter, Smart Design, Samsung, Jawbone, CNN, Philips, Microsoft and was most recently Head of Product Design at Flipboard.
He is currently Assistant Professor of The Practice at the Human-Computer Interaction Institute at Carnegie Mellon University, where he teaches classes in Service Design, Interaction Design, and Designing for AI Products and Services.
New Techniques for Designing for AI
AI innovations have spread across most aspects of people’s lives. Spam filters block messages no one wants, ride-sharing services predict service demand and dynamically adjust pricing, entertainment and retail services recommend desired items and content, bots and robots automate tedious and dangerous work, and intelligent systems help forecast the future, from this week’s weather to the number of sweaters a company will sell to how bad traffic might be. Recent advances created new capabilities, such as systems that detect cancer better than doctors, AI players that beat grandmasters, driverless road and aerial vehicles, and content generation systems that open a world of possibility and raise red flags around ethics and unintended harm.
The success of AI makes it feel like this technology is ripe for innovation. However, today, almost 90% of AI initiatives fail. In addition, innovation teams often fail to recognize low-hanging fruit, situations where a little simple AI would add real customer value. Current technical innovation approaches don’t work well when applied to AI. The HCI research community has been working on how to improve the process from brainstorming to prototyping to delivery. This workshop takes some of what we’ve been teaching our students at the Human-Computer Interaction Institute at Carnegie Mellon University and adopts it for practitioners.
This workshop will be teaching a handful of new techniques that designers, product managers, and researchers can take back and start using immediately. The workshop will be short lectures to introduce a technique, then exercises working with the method hands-on. We’ll start with Matchmaking and ranking: how to find the best uses of different AI technologies. Next, we’ll consider Adaptive UIs: how to find places in our existing products where we can insert AI for the most value for our users. We’ll end with Consequence Scanning to identify risks (and attempt to overcome them).
AI by Design
Most AI projects fail. Some fail quietly before launch; some fail spectacularly publicly, becoming another media horror story about AI. Why does this happen? Because the current process for designing AI products and services is broken, especially when it comes to product strategy—what projects to pursue. But a new approach to designing AI is possible, one that instills more cooperation between designers, PMs, data scientists, and engineers.
This talk walks through a new method that has been developed over many years at Carnegie Mellon University's Human-Computer Interaction Institute. This method uses elements of user-centered design and technology capabilities to find situations where moderate technical performance, high value, and low risk combine to make successful AI projects.
This talk looks specifically about not just how, but also where and when AI should be deployed—and why most companies are doing AI wrong.