Artificial Intelligence in Canada – Laws, Challenges and Solutions
Half a century ago, people would have rubbished the idea of having an “Alexa” at home. But the last few years have made it possible.
Machines are becoming more and more human each day. They are excellent listeners and efficient performers. Once a dream, artificial intelligence has become an indisputable reality in today’s world.
Simply put, Artificial Intelligence can be said to be a technology which is capable of performing tasks which would generally be performed by a mentally and/or physically capable living being such as making sense of spoken language, altering behaviour or solving problems.
The emergence of AI has transformed the way we look at technology. Nonetheless, AI comes with its own share of liabilities and threats. Is Canada equipped to tackle these challenges that lie ahead of us?
Possible Threats/Challenges and Artificial Intelligence –
1. Machine Conditioning – Machines work on the data fed in their systems. A creator or user of a machine can easily enter attributes relating to racism, gender disparity, and religious discrimination. Thus encouraging criminal stereotyping and labelling.
2. Machine Learning – Generally, artificial intelligence works on statistical data and pattern recognition. However, such data may not be relevant while determining the individual attributes of a person. For instance, criminal liability must not be associated with a person only because a majority of his community members were engaged in criminal offences.
3. Hacking and Cyber Security Threats – A cyber attack on AI machine can have dire consequences. It can manipulate the behaviour and decision-making ability of the machines or compromise sensitive data.
4. Increased Unemployment- Advent of AI technology, would reduce human dependency. This would result in lesser job opportunities for humans.
5. Dependency on Power – Relying on machines would eventually make us entirely dependent on power and electricity. A power failure may result in a complete shut down of work.
6. Manipulation of Machines – Since AI decision making is based on a set pattern, some people may devise a mechanism to manipulate these findings. For instance, a serial killer may slightly alter his signature pattern to mislead AI investigative techniques.
7. Violation of Human Rights – Using AI can violate the human right to privacy and life. The encroachment of AI in human lives has raised several concerns relating to privacy and security. The automated decision making by the government in important processes like immigration has attracted severe criticism.
Canada’s position on AI?
AI is stepping into our lives faster than we can imagine. It has entered our homes, police stations, government programs, hospitals, and courtrooms. The Government of Canada has already expressed its intention to work on four core areas of information technology i.e. service, manage, secure and community. It is further stated that the government would use AI in consolation with the administrative law principles such as transparency, accountability, procedural fairness and legality. Further in October 2018, the government released a Declaration On Ethics And Data Protection In Artifical Intelligence (“Declaration”) along with other countries. In the Declaration, the emphasis was laid on digital ethics, fairness, transparency, individual empowerment, mitigation of discrimination, and the creation of common governance principles for AI.
Thereafter, the Canadian government adopted the Directive on Automated Decision Making (“Directive”). The Directive came into force on 1 April 2019 and its compliance is mandatory from 1 April 2020. The Directive’s aim is to reduce risk and increase compliance with Canadian Law. In order to ensure transparency notice before decisions and reasoned orders have become mandatory.
The government would determine the appropriate license for software components. The Automated Decision System(s) can be reviewed by the government or government authorized third-party agencies. The release of custom source code owned by the government unless it is exempted under the Directive.
Moreover, there are specific guidelines incorporated in the Directive pertaining to quality assurance. The Automated Decision System(s) must be tested before launching and monitored as per scheduled timelines. Validation of data, employee training, peer review, establishing contingency systems, conducting risk assessments, legal compliance and allowing human intervention is incorporated to assure quality.
The agents of Parliament are excluded from the application of the Directive. The Treasury Board needs to facilitate the implementation of the Directive. Non-compliance with the directive may attract any punishment deemed appropriate by the Treasury Board.
Conclusion – The Road Ahead
The Declaration and the Directive are vital steps taken by the government to control the imminent challenges posed by AI. The contingency plans to offer back-up plans, human intervention to control bias and validation of data are few of them.
However, some of the issues still remain unaddressed. The government must issue a list of sectors where the use of AI must be restricted or prohibited. In France, predictive litigation AI has been prohibited. Similarly, many countries are opposing the use of fully autonomous weapons. It is essential our government adopts a clear approach towards such sectors.
The non-applicability of Directive on agents of Parliament can pose certain issues. For Instance, the Chief Electoral Officer of Canada (CEO) and The Office of the Privacy Commissioner discharge essential functions and use of unregulated AI by such offices may prove dangerous in the future. Specific penal provisions must be enacted to ensure compliance of the Directive. In my opinion, there must be separate legislation for the use of AI in private enterprises.
AI is no longer the future but has become our present. Unless a clear approach is established and adopted by the government, these threats cannot be adequately handled.