The big tech companies are all working diligently to optimize artificial intelligence to better serve their customers’ searching needs. Better customer experience is known to directly impact adoption rates/conversion rates and create paths to other revenue streams. But many of the optimizations and data mining efforts for improved machine learning and intuitiveness are coming at the expense of privacy for the individual. Here’s how:
More Audio Eavesdropping
It’s already no surprise that many tech companies have come under fire for eavesdropping and transcribing phone and messaging app conversations for their pursuit of improving AI’s comprehension of spoken language. Much of the information they gather for human review is anonymized, and is typically recorded only after the device wake-word (“hey Siri,” “Alexa,” “Cortana,” “Amazon,” “Echo,” “okay Google,” and growing) has been spoken, but it can be triggered simply from a misunderstanding in casual conversation or sitting on your phone’s home button.
And in listening to these conversations, confidential details can quickly make a person much less anonymous. Google, Apple, and others have made it a point to create security safeguards, but these often come in the aftermath of leaked private information and device infiltration. Nothing you speak can truly be secured from abuse, outside of disconnecting entirely, especially when the Terms and Conditions articulate “you give [us] (and those we work with) a worldwide license to use, host, store, reproduce, modify, create derivative works or other changes we make so that your content works better with our Services.” The best one can hope for is limiting the amount of information shared via your device(s) in its feature settings.
The Apple iPhone at default will let Siri observe browsing behavior regardless of what browser settings you have deployed. A switch for it is provided.
Many apps can share location details in the background to work better with the app interfaces. You should be cognizant of whether it would be crucial they do “always,” “never,” or just “while using.”
Any search activity (especially when logged into a Google profile) is cataloged and associated with the machine’s IP address and geographic region the search was performed from to provide custom results. This data sharing and historical activity retained can continually inform further customized results from work, on the go, or at home from any related device. To limit to a degree Google device history and data sharing, go to https://myaccount.google.com/activitycontrols and pause certain options.
Another simple method to restrict sharing sensitive search and location information can be by deploying a VPN to web-connected devices.
Behavioral and Facial Bio Feedback
Back in 2015, Google/Alphabet filed a patent to retrieve the users’ “emotional state” from mobile and tablet devices to determine and obtain a “first emotion classification tag.” This would essentially become a sort of biometric ranking factor to correlate with a satisfactory search result page or visited website versus an unsatisfactory one. Emotional aspects that would naturally be sensed through the forward-facing camera are –
- Blinking/blink rates
- Pupil dilation
- Squints and facial twitches
- Face blushing/flushing
- Thermal heat mapping of face
- Resting versus active breathing
Google/Alphabet on top of that, and along with their acquisition of camera sensory and surveillance technology company Nest Labs, filed another patent for “occupancy sensor[s]” to detect the emotional and environmental data related to behaviors, activities and expressions of smart home occupants to record for subsequent use with smart devices. This sort of opt-in surveillance, with the previous track records of malicious device hacking or human mishandling of recorded data, should give one pause before integrating smart devices into each location of your home for the sake of receiving a more tailored digital search experience. As of now, there is no evidence that facial biofeedback has been integrated into any search algorithm programming nor any intrusive data mining coming from smart home devices, but there is the potential for it being explored (and opted into accidentally), and it is possible we won’t discover it’s happening until after it’s been happening.
Evolving from Traditional Search to More Predictive Search
Auto-suggestions and travel pattern–tracking in Apple Maps are just a few predictive technologies that have already been adopted into the status quo of the digital world. These technologies load popular searches and frequented points of interest more easily.
It can be expected that other data points can eventually be incorporated into various applications built around your activities and other feedback that amounts to a pseudo-predictable and habit-formed profile that artificial intelligence grows for each specific user, and tailors their services to. The goal – to serve information for a specific user instantly in order to make it (a) more difficult to game their algorithm and (b) assist the user with as little required effort on their part as possible.
Eventually, Amazon, Google, or other AIs will know more than we realize, such as, when a certain restless sleeper awakens, whether they reach for an eBook, take a sleeping medication, activate a white noise-making device, or schedule an appointment with a doctor in the morning. When those behaviors are learned, search results may not even be required.
A doctor’s schedule availability, a new white noise sound recommended from YouTube, or the latest releases from your favorite book genre will appear on devices without a search being conducted. Gathering these specific data points at the personal level for every individual to suit their lifestyle and needs can have tremendous value for both the consumer as well as the business service provider, but it will no doubt come with certain costs.