Advertisement
Artificial intelligence
OpinionComment
Andy Chun
Andy Chun

Opinion | When it comes to using AI, the real question is one of trust

  • Mechanisms need to be put in place to ensure users can trust recommendations, predictions and decisions made by AI systems. AI needs to be explainable rather than a black hole.

Reading Time:3 minutes
Why you can trust SCMP
0
Some people use facial recognition daily to unlock phones. Photo: Shutterstock
Andy Chun
The topic of facial recognition has stirred up quite a lot of controversy in recent years. There are growing concerns, particularly over data privacy and data protection, especially now that such technology is so prevalent. Some people use facial recognition daily to unlock phones or authorise payments, and many encounter it in security at workplaces or airports.
However, people have started to question the ethics as well as the accuracy and potential bias of facial recognition technology. In a recent study by the Massachusetts Institute of Technology, it was found that the technology can be gender- and/or race-biased, probably due to imbalances in the training data used. In another study last year, 28 US Congress members were mistakenly identified as criminals. Some cities in the US have even started to ban the use of the technology, including by the police. British MPs believe legislation should be put in place before facial recognition trials start, while the European Union is seeking regulations to limit what it calls the “indiscriminate” use of facial recognition.

These events are not surprising for any new technology with such a wide potential impact on our lives. Careful attention and scrutiny should be paid to the accuracy of the technology, as well as to informed consent, legal liabilities and ethical use. These are all healthy steps towards ensuring legitimate use while respecting the rights of individuals.

But artificial intelligence and deep-learning technology need not be privacy invasion tools.

The AI technology behind facial recognition has much wider uses. It can be used not only to identify a person, but also to detect the presence of a face and its features, as well as to predict health, emotions and mental states. Depending on the application and what algorithms are used, the data privacy implications may be different.

Recognising faces is also only one of the many important features deep learning technology has to offer. The most interesting is probably facial analytics, which analyse photos or videos to extract facial characteristics and predict gender, age and even body mass index. It has been used to detect early signs of diabetes, heart disease or dementia, and in measuring heart rates and blood pressure.

Advertisement