Hidden Artificial Intelligence

Artificial intelligence (AI) is an integral technology of modern society. Whether it’s online streaming platforms, smartphone assistants, or mass surveillance, AI has long established its presence in various facets of daily life. Some of these technological advancements have been well-documented, while others have quietly become influential to the point of global mass adoption.

We’ve put together a list of five AI programs we use everyday – programs that have been under the radar for the general public or, at the very least, accepted without too much thought.

Face recognition

Do you know those little squares that pick up faces on your phone’s camera? AI drives that software taught to recognize human faces. Apple’s FaceID uses similar tech. Smart airport boarding gates are able to screen passengers via facial identification and risk assessment.

In Paris at Charles de Gaulle airport, ‘smart gates’ have made managing millions of annual tourists a lot easier. DeepFace, a neural network in development by a research group at Facebook, is said to have a higher rate of facial recognition than the FBI’s own Next Generation Identification system.

Apple and Facebook aren’t the only tech giants with powerful recognition software. Biometric ID verification has been used by Microsoft for their Kinect peripheral on the Xbox home console, as well as Windows 10’s infrared verification system.

Healthcare AI

From drug research to robot-assisted surgery, AI is already in use for a variety of medical scenarios. The majority of these applications are data-centric, concentrating on statistical analyses and improvements to diagnosis accuracy. There are several more sophisticated implementations of AI as well, such as robot-assisted surgery and medical research algorithms.

Many of us own FitBits, Smartwatches, and other wearable health devices, and all of these gadgets use AI to keep track of our vitals. In Massachusetts, a machine learning AI is assisting with patient diagnosis. Known as PathAI, the system’s goal is to reduce errors in medical diagnoses.

Harvard Medical School uses a chatbot named Buoy Health to assess a patient’s health and advise them on seeking assistance for any health issue. Enlitic is a deep learning machine used in California as a means of speeding up radiology work. According to engineers at MIT, Enlitic is one of the five most advanced AIs globally, ranked above Google’s prolific AI systems.

Online communication

Without doing any research, would you be able to say which year Google began offering Smart Reply as an option for your emails? Or did you merely note the convenience, without so much as a second thought? You wouldn’t be alone in that experience. Millions of daily users wouldn’t be able to tell you that Gmail’s tactful AI addition was released more than four years ago.

In order to infer the context of an email, Google’s AI can understand nuances of language like tone, inflection, and turn-of-phrase. A similar implementation is used for Gmail’s spam filter. Machine learning and natural language processing (NLP) sift through our email accounts on a daily basis, filtering out ads and scam mail with uncanny accuracy on a truly massive scale.

Autocorrect, and text editors can predict what you want to say before you’ve typed it. Chatbots can recognize speech patterns and adjust their responses accordingly. Those pop-up chats from ‘service representatives’ on various websites? They’re all chatbots. Are the ‘doctors’ ready to give you instant advice 24/7? Also chatbots. Even Reddit is positively flooded with bots.

Banks use AI to facilitate the transfer of payments. AI is used to monitor for fraudulent activity on a wide range of accounts and can pick up irregular expenditure via machine learning. As transactional AI continues to develop, credit scores and loan reviews increase accuracy, verification services get streamlined, and identity management is becoming an automated process.

Search recommendations

Predictive search algorithms trawl our entire viewing history in order to suggest recommendations on what to watch or listen to next. We’ve gotten so used to these AI suggestions that many of us overlook the fact that our exploration of multiple new genres is severely hampered by these algorithms.

Netflix is an excellent example of predictive tech at work. The streaming network can gauge a viewer’s taste in movies and TV shows, and the accuracy of these recommendations is uncanny. The downside to this AI is that smaller shows and lesser-known films don’t show up at all, leaving a vast portion of content in the dark for most viewers.

Ridesharing services like Uber and Bolt have been using AI since their inception. Every detail is worked out by computer software, including GPS tracking, wait times, and fee structure. Machine learning is used to control ‘surge pricing’ (i.e., peak traffic times). Fraud detection is also overseen by AI, as are meal delivery times and optimal pickup or drop-off locations.

Smart recommendations are present in various other forms of media. Valve’s video game storefront, Steam, uses data analytics to curate a list of game titles the user is likely to be interested in. Facebook feeds are curated to a point bordering on manipulation, and it’s all based on AI-powered predictions.

digital trasformation

Travel & navigation

AI has already influenced transport and vehicle industries worldwide. Self-driving cars have hit the mass market thanks to companies like Tesla and Google. The US and China recently made driverless taxis open to the public. Smart cars can adjust suspension, power output and even help to prevent collisions or accidents.

Your smartphone’s GPS has a built-in AI. Through the use of location-based services, your mobile can monitor traffic flow and provide you with the most efficient route for your trip. The AI learns about how, when, and where you drive so that it can give accurate feedback based on your individual needs. Tracking technology has also been used to significant effect as an anti-theft tool.

There are five levels of autonomy that dictate the class and capabilities of any given car, all of which are based on the amount of human intervention required. These levels range from basic advisory AI at level zero to complete autonomy by level five. At present, all five levels are already used in a number of real-world scenarios.