UX designers face a world of intelligent machine, AI. We don’t need to know everything about the AI. But, intelligent technology always demands good design. Now is the time to reconsider the intersection of machine intelligence and thoughtful, inclusive UX design.
Machine Learning for Designers
Take Machine Learning. Basically, machine learning means the construction of algorithms that can learn from and make predictions on data. It creates evolving feedback loops over time for more sophisticated prediction model.
Collecting human feedback and data is crucial in achieving intelligence. UX designers should help making this loop to be more useful to users. The question then becomes; how we as UX designers can make more meaningful relationship that engages users to the AI? We applied this question to travel planning experiences that provide personalized recommendations.
Design for Embracing Personalization
“We’re entering a new era that embraces personality, rather than minimal perfection” — Uurasjarvi
Minimalism has been popular for a long time. That was fine for commoditised mobile apps and in most cases, it still is. However, we’re currently witnessing that minimalism isn’t powerful to support personalization. Minimal designs cannot offer as variety of opportunities to delight users and differentiate AI supportive features. So, the new way of design that can serve personalisation comes : Maximalism.
Maximalism can create viable and distinguished expressions. It can enhance experiences across a variety of personalization that are normally overlooked or simply combined under a common minimal design. We can design more heightened expressive AI to keep us from having ordinary experiences.
Maximalism with Random Effect
Let’s move to interaction. Building a give-and-take relationship between the AI and users is one of the biggest design challenges. To facilitate successful exchanges, we focus on the design improving the AI and keeping user interested by operant conditioning, where an antecedent stimuli is followed by a consequence of the behaviour through creating a reward (reinforcement).
B.F. Skinner in the 1950s suggests using unpredictable rewards called variable ratio model to keep behavior going and establish a habit. Skinner observed that lab mice responded well to random rewards. As he discovered over 50 years ago, variable rewards are a powerful inducement for creating habits in technology-induced behaviors.
Consider random variable effect in this relationship. It can dramatically alter the nature of conventional prebuilt UX flows. Less monotonous experiences with random feedback can make AI interaction more enjoyable and engaged. Besides, induced feedback loop help with machine learning by gaining the data AI needs to get smarter and be more relevant.
As intelligent systems move forwards, it’s not easy to anticipate users’ adjustable expectations. Users’ preference can change over time, and a machine should keep following this. There will be a gap between users’ expectation and AI’s prediction all the time.
Bob Dylan said, “To live outside the law, you have to be honest”. The AI have to be honest to live with user sustainably. Since it’s not possible to have perfect prediction, we can give snippets of AI’s statistical correlations behind the scenes. Recommendation with apparent sense of information will make the AI trustworthy.
To be forward
We strongly believe there will be a rich future in design of machine learning. We are just in the beginning stages of the journey to gaining a better practice and knowledge in this area.
In this article, we approached the interaction of AI theoretically with core values of Maximalism, Random Effect and Trustworthiness in travel planning experiences. They are still hypothetical assumptions to be proved yet.
There’s always going to be room for problem solving when humans interact with technology. If UX designers are willing to experiment and question about machine learning, we will meet this challenge in creative way.