Digital Behavioral Design 37: AI Nudgebots & Adaptive Tech ??
Dear Friend.
It's been a busy half year, with back-to-back training. One topic that keeps emerging is the dream of automating behavioral science with AI.?
This topic is at the center of my passions, the fusion of predictive statistics, neuroscience, and behavioral design.?
AI will drive the next generation of digital products. They'll anticipate users' needs and send the right message at the right time. Moreover, they'll act like a charming host, making users feel welcomed, supported, and respected.?
Technologies that understand their users better have obvious advantages.?
For instance, I won't use Apple Music for free and prefer to endure ads on Spotify. Why? Spotify has competent data scientists. Apple has incompetent data scientists. Spotify keeps finding new music that I love. Apple is nowhere close.?
Spotify's algorithms know me better, making it a better product--at least from my perspective. Conversely, Apple Music wastes my time with sub-standard UXD, and crappy algorithms.?
If you're in a relationship and your partner cannot figure out what you like and dislike, something's wrong. The same goes for digital products.?
The next generation of psychology-driven products will read users' intentions and adapt to their needs and preferences. It comes down to tailoring--delivering the most promising content based on a person's background and current desires.?
Eventually, users will feel angry at products that fail to read their minds. Sending the wrong message at the wrong time is annoying. Relevance and timing are everything.?
Customer satisfaction is perhaps the best "off-label" way to use behavioral science. But we have a problem. There are three steps involved in building adaptive technology. Unfortunately, there are barriers in each.?
First, it's hard to get our hands on good data. Assuming we're respecting opt-ins and privacy laws, you'll rarely have access to viable user data unless you've built your platform.?
Second, assuming you can access the data, can you build an algorithm that predicts users' preferences and current emotions? And if so, what theoretical framework are you going to use. There are many competing models, some back by science and others as junk frameworks that psychologists have debunked. Shopping around for valid psychometrics and behavioral frameworks requires a rare combination of skills.?
Third, let's assume that you can tailor every product page to your users' state of mind. How would you use this insight to build great user experiences? How would you use this to boost sales? How would you use this to help your users achieve the goals that matter to them?
It's here that AI and behavioral science work together. AI provides insights, while behavioral science gives us the models and techniques that help us find the optimal messaging for each situation.?
Eventually, AI will take over this step, but for now, we humans are still in charge.
Here are some related articles.?
Best,
Brian
ARTICLES
领英推荐
How Netflix’s Choice Engine Drives Its Business
Netflix’s existence depends on finding programs that you will want to watch from its library of content. In this article, Eric Johnson analyses Netflix from a choice architecture perspective. This is just one of many perspectives, but it's a good teardown from a BE point of view.?
A/B Testing Website Copy With GPT-3
AI is now making suggestions on how to author higher-converting copy. We're still in the early days, but check out the promising results by VWO's recent dive into the GPT-3 text authoring AI.??
Human vs. AI competition
Who writes higher-converting copy: humans or AI? See the results of VWO's contest.?
Math Equation of Mood Can Predict Your Preferences
Can we use math to measure how users feel about our products? Potentially, with a "hedonometer", which scientists are developing to read how we truly feel about things.?
Wishing you the best.
Brian
==================
Brian Cugelman, PhD
Behavioral Design Academy of AlterSpark