Automated facial recognition technology is rapidly spreading throughout Australia, with various sectors such as retail, sports stadiums, and casinos already utilizing it. The Australian government’s digital identification system is also set to expand in November, following the passing of new laws earlier this year.
As the technology becomes more affordable and advanced, it opens up possibilities for a wider range of applications, including an age estimation tool. To gauge public opinion on this rapidly growing technology, a team of researchers conducted a national survey of over 2,000 Australians. The results, recently released, indicate a general lack of knowledge about the technology and a range of attitudes towards it.
These findings are crucial for policymakers in order to maximize the benefits of facial recognition technology while minimizing its potential harms.
Facial recognition technology operates in two main ways. The first is known as one-to-one use, which verifies a person’s identity, such as unlocking a smartphone. This can greatly enhance convenience, as individuals would no longer need to carry multiple forms of identification and can simply undergo a face scan.
On the other hand, one-to-many uses of the technology enable the identification of unknown suspects or individuals in a crowd. In both cases, the technology creates a template from a known individual’s photograph, which can then be compared to new photos to determine a match. However, the technology provides a probability of a match rather than a definitive yes or no, and it can still be deceived by masks or disguises, though its ability to overcome these challenges is improving.
Attitudes towards facial recognition technology vary depending on its specific use. The survey revealed that nearly three-quarters of Australians have little knowledge about the technology, with only 5% claiming to know a lot about it. The majority of respondents expressed support for one-to-one uses, such as accessing government services (57%), identifying criminal suspects (75.2%), and verifying identities during disasters or war (80%).
However, there was less support for other uses, such as workplace tracking of employees’ locations or targeting shoppers. Respondents strongly believed that facial recognition technology should not be used for commercial gain.
Perceptions of the accuracy of facial recognition technology were generally positive among survey respondents, with most considering it accurate or very accurate. However, it is important to note that different systems have varying levels of accuracy, and certain demographic groups may experience lower accuracy rates, raising concerns about racial bias and the potential misidentification of innocent individuals.
Transparency and consent were crucial factors for survey respondents. 90% of Australians expressed a desire to be informed about when and where the technology is being used on them and to have the opportunity to consent to its use.
Governments should take public opinion into account and respond accordingly. Automated facial recognition technology is a powerful surveillance tool that raises significant questions about privacy and civil liberties. In 2019, a proposal for a national face recognition database for law enforcement was deferred due to concerns about public response and enrollment in digital ID programs. Recent legislation has restricted one-to-many matching using the national facial recognition database, but individual states still maintain their own databases. Concerns also exist regarding the government and private sector’s ability to securely store and manage people’s data, especially considering recent data breaches.
While automated facial recognition technology can be beneficial, there is an urgent need for better public education on the technology and the associated issues to ensure responsible and democratic use. Legislation dedicated to minimizing the risks of creating an automated surveillance society is also necessary, as argued by former Australian Human Rights Commissioner Edward Santow.
