It's about to get personal — even more personal, that is.
We're already accustomed to AI-driven personalization of everything from book and movie recommendations on Amazon and Netflix to music playlists on Spotify.
Banking, finance and investments
Business and industry sectors
Business, economy and trade
Computer science and information technology
Diet, nutrition and fitness
Economy and economic indicators
Exercise and fitness
Health and medical
Software and applications
But 2019 should bring what we could consider "hyper-personalization" to consumers through a range of applications focused on health, finances, shopping and everything in between.
This is due to the growing capabilities of AI applications to make more accurate and cost-effective predictions, the ever-larger pool of personal data from which such applications can draw, and our collective willingness to "opt-in" to customized suggestions and services.
Here are some areas in which we can expect AI-based hyper-personalization in 2019:
Physical health has become a hotbed for AI applications.
Fitness apps, and wearables like the Fitbit and Apple Watch, and even some VR headsets can automatically collect people's data — such as how many steps and flights of stairs you've climbed — and aggregate these across the user population to generate even more personalized fitness profiles and plans, much like the Nest smart thermostat uses data from across homes to optimize energy-use plans. Suggestions like "Reduce your food intake 210 calories/day to increase your life expectancy by 1.5 years" may not be far off. Fitbit and Google have already teamed up to combine personal fitness data with medical records to deliver a more complete health care profile to physicians.
The problem to date has been that most fitness app/wearables users tire of entering the foods they've eaten — or cheat and choose not to enter them. That results in the "garbage-in, garbage-out" problem: Without accurate data, analyses are meaningless. AI has an answer for that: As AI becomes more advanced, you'll simply be able to take a picture of what you eat, or the app will monitor food intake through your smart glasses — automating the data collection process. Some apps, like Lose It, are already using this kind of technology. No more "forgetting" to log the cheeseburger and fries you had for lunch!
Mental health is another rising domain of AI application.
Want to understand what times of day you feel most down, or what people bring out your highest and lowest energy levels? There will soon be an app for that. Again, the idea is to cut out the middle person: you. When we're in charge of telling a therapist how we feel or entering it into an app, it's easy for our biases to influence what we share. Apps will be able to do the sensing on their own, using biosensors to monitor heart/breathing rate, galvanic skin response, and other indicators to gauge our reaction to everything from the morning commute to dinner with the in-laws. Then we can use the comprehensive reports created to fashion lower-stress lifestyles, or at least be aware of the people and situations more likely to make our palms sweat.
Quartet Health, for example, harnesses machine learning to look for signs of mental health issues in users. When it detects likely problems, it connects users to practitioners including therapists and doctors for treatment. Empatica offers a wearable that detects physical signs of imminent epileptic seizures and alerts users and caregivers; the company plans to move into more general health applications. The Mindstrong Health app monitors a wide array of health indicators to move from "reactive to proactive."
AI-based facial recognition technology also has broad applications in mental health. Affectiva, a business spun out of MIT, offers products that "emotion-enable" apps by reading and analyzing facial expressions with better accuracy than humans do. So your apps should soon be able to gauge your level of stress or anxiety/depression symptoms and make suggestions accordingly: "Do two minutes of deep breathing" or "Make appointment with therapist." The company has also developed an in-car product to assess emotional state while driving from facial expression and voice, improving road safety.
Personal finance is increasingly a DIY, or SDI (self-directed investment), domain. But how do we know which investments best match our highly individual growth, risk and liquidity needs and tolerances? And when to invest in them? And how much to invest? And when to sell?
Short answer: We can't. Not easily, anyway.
AI can help. Applications will increasingly be able to predict our risk appetites and other investment-profile features by literally "reading between the lines." For example, at my lab at the Northwestern Institute on Complex Systems, we have developed a deep-learning algorithm that can analyze verbal or written text to assess the speaker's/writer's level of confidence in a given statement on anything from how effective a new drug compound might be to the accuracy of a company's stated earnings projections.
A personal finance app incorporating such technology could ask you to write responses to several different risk-tolerance scenarios, then assess how confident you really are about each to generate a highly personalized investment plan for the short or long term. In fact, the early-stage firm Narrative Science is using natural language processing to develop applications of this type, and could potentially help users understand their tolerance for investment classes such as small-cap stocks.
As shoppers, we want to be very individualized in our choices, from apparel to eyeglasses to automobiles; but we also want to buy things that others like us, or those we want to be like, find appealing. Retail-focused AI applications can help determine your product preferences by looking not only at your past purchase behaviors, but those of others like you, to generate recommendations based effectively on a form of behind-the-scenes crowdsourcing.
Of course, some of this is happening already. It's safe to assume Amazon has been aggregating personal data to develop product recommendations and other predictions for some time now. But the scale on which this will happen in 2019 — across industries and types of products — and the level of personalization to expect should both grow. Elemental Cognition, for example, a company launched by the former leader of IBM's Watson team, is developing AI-based technology to help us understand why we want what we do, with application to shopping, personal finance and other areas.
In the bricks-and-mortar world, enhanced facial recognition can mean a much more personalized in-store experience. Retailers, for example, would be able to recognize loyalty-program customers who've walked in — and opted in for recognition — and push personalized deals and recommendations to them through their phones, along with offering a higher level of service (such as a shopping consultant). Facial recognition would also enable the business to analyze a given customer's or segment's shopping style or in-store path more closely.
In general, AI-based hyper-personalization offers large potential benefits, but it also warrants a word of caution. The more data we place into the hands of the businesses increasingly crowding this space, the more potential risk of misuse and theft. In short, the more of ourselves we trust to the cloud, the greater the need for failsafe cybersecurity. AI applications will most likely help with that before long too.