Swiss think tank weighs in on biometrics: AI is good – on a very short leash

A European organization formed to search for the benefits and risks of new technology has issued a report urging Swiss legislators to ban three uses of speech and voice recognition algorithms and put a moratorium on a fourth.

The recommendations are made in a report by the publicly funded Foundation for Technology Assessment.

(An English translated executive summary begins on page 22, and an English-language site has been created to explain the foundation project.)

The foundation, also known as TA-Swiss, contends that too little attention is being paid by Swiss legislators to the negative impacts of systems as basic and common as Amazon’s Alexa digital assistant.

A ban should be placed on autonomous, real-time government surveillance networks as well as any government-run social credit schemes that use biometric algorithms involving speech, speaker or face scanning, the report’s authors assert.

Autonomous decision-making should be banned as well if a system uses speech, speaker and face recognition in “important life domains.” Examples including health, finance, employment and law enforcement. Semi-autonomous algorithms should be closely monitored by trained staff who are responsible for the decisions.

Also worth banning, according to the TA-Swiss report, are smart glasses worn in public. Grouped in this topic are other hard-to-spot technology that can surveil public spaces.

A moratorium is need on algorithms used in “sensitive life domains” to detect emotions and disease until they are proven reliable.

While the authors were at it, they recommend that speaker detection, specifically, and biometrics, generally, should not be the only bases for authentication.

The report suggests lawmakers create infrastructure capable of educating and protecting the public from abuses that harm civil rights.

High-risk tools should be regulated, for instance. Legal rationales should be established for public deployment of algorithm-based tools. Education and training should be offered for those using biometric recognition tools.

It also encourages supporting people who are subject to data collection and better enforcement of their rights. A national conversation about what biometrics can offer as well as threaten should be encouraged.

And, last, the authors want “sufficient resources” given to trusted organizations working to inform and support people living with biometric surveillance.

Article Topics

AI  |  algorithms  |  biometrics  |  consumer electronics  |  legislation  |  responsible AI  |  surveillance  |  voice biometrics