How much data is there
Human beings generate a lot of data. How much? On average each person on the planet creates 1.7mb of data per second. That’s about the size of a decent quality selfie taken from a smart phone, or a really large spreadsheet. Pick your poison. According to Domo, a business intelligence company, humans collectively created 2.5 quintillion bytes of data in 2020. I’ll save you the Google search, that’s a number with 18 zeroes. It’s huge. For a comparison the number of drops in all of Earth’s oceans is a number with 26 zeroes. It’s conceivable that at our current rate, in the next 20 years we’ll have as many selfies as there are drops of water in the ocean.
That’s so much data that if all ~7.9 billion of us humans decided to band together to consume it all, it would still take us thousands of years. We used technology to make that ocean of data and the only way we can conceivably make sense of it in our lifetimes, is through technology.
Big tech (Google, Apple, Amazon, Microsoft, and Facebook) is helping us generate all this data and typically they’re profiting from it. The question I want to answer is how can society-at-large profit or at a minimum benefit personally from all this data. When the day comes that all of the data big tech companies have been collecting on us is turned over, what are we going to do with it?
What can we do with the data
One obvious solution is mining that data on behalf of the public. Ten years ago I created an app that sought to data mine public video surveillance footage. Admittedly that idea was solving a problem many people didn’t know they had. It was also too soon for the technology. Today is different. We now have the affordable cloud computing and advanced technology to manage and mine data, on-demand for our personal benefit. The results of this data mining won’t be business charts and graphs. They‘ll be the answers we get in conversations with personalized autonoumous agents.
When I say agents the term I should really use is bots. Not the fun, clanky, anthropomorphic robots like Bender or C3-P0 immortalized in pop culture. I’m talking about the bots crawling the web to help us search better, swarming computer networks in orchestrated attacks, and even most recently helping scalpers buy up limited edition sneakers or grabbing coveted next gen game consoles before anyone else. When you call a customer service number, it’s likely a bot that answers the phone. Chatbots have been around forever and since AOL’s SmarterChild back in 2001 have shown increasingly impressive conversational skills. Thanks to advances in machine learning for language, chatbots are now so advanced and convincing that people are getting into arguments with them on Twitter.
Recent advances have made bots, like AI personas able to be personalized by more human attributes like our voices and even speaking style. There’s been an AI created to generate rap lyrics similar to Kanye West and The NY Times recently ran a story lamenting the “digital reincarnation” of Anthony Bordain into an AI that sounds like him. The merging of personal data with conversational AI to produce really smart, personalized assistant bots is aleady at our doorsteps.
What data should these smart bots start with
The need for a bot personalized with my health data hit home when I caught covid-19 in March of 2020. Afterwards despite a very busy life, I got serious about improving my cardio-vascular health while also trying to manage and track my diet, sleep, medications and fitness. I assumed there was some magical app to help me do this. I asked my friends, family and even healthcare professionals what they recommended but there were no stand-out solutions. According to Statista, there’s an estimated 50,000 mobile health apps, and Americans on average delete them within the first 7 days of downloading for reasons including privacy concerns, cost, ease of use and accuracy.
I built Roqo (Robotic open-ended query operator) with the ability to help me track my data. Data like my fitness, blood pressure, diet, sleep, hydration, medications. I could simply text Roqo my questions and get the answers back in plain language based on my data. No charts or medical jargon. I got so used to having conversations with Roqo that even when I could have just searched the web for a generalized answer, I defaulted to the chat because it seemed like more intuitive, accessible and private experience. In addition just by asking a question like “how many carbs did I eat today from a chicken burrito, carrot juice, blueberry scone and caesar salad?” The bot could then provide the option to log those foods as part of your food journal.
Health insurance companies, healthcare providers, start-ups, and of course big tech are all rushing to provide services in this space. These businesses see the potential for huge profit in giving our medical data over to AIs. Most of their solutions appear to be centered around diagnostics and remote monitoring for sick patients. I think that for this space to innovate it should be focused on the healthy first. It’s the healthy that drives the wider adoption of new technologies and trends like the Apple watch, Peloton, and uh the Keto diet.
Everyday data isn’t healthcare data
We’re all so busy managing our lives that we can barely remember how much water we drank yesterday or what we ate last week. Most of us will never write these things down. What if a person were trying to cut down on their sodium intake for the week, but has no idea how much sodium they’ve already eaten. A bot like Roqo could give them that information and even advise them whether or not getting take-out from their usual spot is a good idea this week.
Bots capturing and processing this everyday data could give us reminders and potentially open up entirely new insights into managing our lifestyles.
I’ve been wearing an Apple watch almost everyday for over five years and it’s an important tool in motivating me to exercise. Understanding 5 years of fitness data could have helped me know a baseline for how fit I was prior to covid-19. My doctor doesn’t want 5 years of fitness data because my doctor isn’t a data scientist. Apple provides some charts like resting heart rate and V02max but does the average Apple watch wearer have any idea what to do with that? Not only could a bot process years of fitness data, but it could also give you answers and tips to improve your fitness via chat.
Most healthcare data is created from some form of diagnostic tests. These tests are conducted by licensed professionals in offices, labs and hospitals. It’s extremely useful data to be sure but unless you’re always sick you won’t collect diagnostic data very often. Our everyday health data might tell us as much if not more than the couple of yearly check-ups at a doctor’s office. Daily metrics like what we eat, how much we sleep, how much we exercise, or how consistently we take medications and supplements are invaluable in understanding our overall health. We can’t learn from that data if we don’t consistently collect it, and in 2021 that’s still too difficult for the average person. I envision a near future where everyone is chatting up their personal bots to enjoy a healthier lifestyle.
Kaza Razat is an AI Technologist and creator of Roqo, an AI powered health and wellness chatbot. Kaza writes about AI, human augmentation and creative uses for advanced technology.