Globally, the average adult consumes the equivalent of about one bottle of wine per week. That may seem a reasonable rate, but many people can’t stop there. Serious health and social problems emerge around binge or excessive drinking: illness, crime, traffic accidents, addiction, etc. Alcohol is a factor in 5.3 percent of all deaths worldwide.
There is increasing attention on machine learning, deep learning, IoT and computer vision technologies in attempts to reduce the damage done by alcohol and improve the safety of drinkers. AI-powered models can ensure alcohol purity, preventively monitor and assess human behaviours related to drinking, and generate support and services for addicted, intoxicated or unconsciousness people.
Case Studies
Moufans Commune – Maotai Authenticity Identification APP
For generations, Chinese Moatai has been a deeply-loved, top-level distilled liquor. It has also been plagued by imitations that disrupt the market order. Changes in packaging and a large number of limited editions have also made it difficult to identify genuine products.
To protect consumers from potentially dangerous knock-offs, the Beijing Maotai Culture Research Association and Moufans Commune introduced their Maoyou Gongshe App, which can confirm authenticity using computer vision, object recognition, deep learning techniques and convolutional neural networks. The App uses picture-segmented patch sets to train convolutional neural networks. It can increase training samples and improve the recognition rate during occlusion. The overall recognition rate of this method can reach 97 percent, and it has strong anti-interference ability and robustness.
TiSPY – Alcohol Checker
The TiSPY personal alcohol gadget has a built-in alcohol concentration sensor system, which displays current alcohol blood alcohol content when users blow into it. Its machine learning function can predict when the user is likely to reach a state of drunkenness and warn the user to “avoid hangovers”.
TISPY has Android and iOS apps that wirelessly sync data with the device to display users’ drinking state and habits.
Uber – User State Predicting System
Uber is working on AI technologies that can be used to detect whether users are drunk. The user state predicting system uses machine learning, IoT and deep learning techniques including decision trees, support vector machine (SVM), sensor technology and neural networks to identify potential drunkenness through subtle changes in users’ behaviour while using the App.
The system uses data gained through past trips to train the computer model. These include pick-up locations, walking speed, unusual typos made when entering a ride request, the angle at which the user holds the phone, and whether the phone is swaying, etc.
If the system flags an unusual ride request, it will adjust Uber’s service accordingly. Possible solutions include directing users to pick-up points that are better lit, or matching users with drivers who are experienced with drunk passengers. The system could also prevent drunk users from sharing rides with other passengers.
What’s Next?
Because humans’ self-protection instinct is lowered by alcohol, artificial intelligence has great potential for development and deployment in drunk protection. More practical tests will however be needed to ensure the stability of the apps and various products and especially their safety for users.
It is expected that increasing public awareness of the dangers of alcohol and the steady development of relevant AI-powered technologies will land more and applications aimed at risk-reduction around drinking.
Author: Xuehan Wang | Editor: Michael Sarazen
