Submissions to the House of Commons Standing Committee on Human Resources, Skills and Social Development and the Status of Persons with Disabilities

The caselaw that interprets section 8 of the Charter has recognized that we have an elevated expectation of privacy in our homes, owing to the private, personal, and intimate nature of the activities that take place there.21F Accordingly, information about activities inside the home is afforded a high level of constitutional protection when and to the extent that it can reveal intimate, personal i. [...] If discriminatory bias is present in the dataset that the AI model is trained on – whether in the underlying data itself or in the way the data has been packaged and formatted to render it intelligible to the program – the AI will replicate this pattern. [...] There is also a risk that bias will be introduced if an AI model is used for a different purpose than it was trained for, as the new context may raise forms or expressions of bias that were not considered or controlled for in the preparation of the training dataset.98F The decision-making and analysis of AI models are often opaque, functioning as a proverbial ‘black box’: the inputs and outputs ar. [...] Without examining the code and data that produced the model, bias can only be evaluated by examining the outputs of the AI model, a process which requires not only a substantial sample of outputs but also technical and statistical expertise. [...] Summary The BC Civil Liberties Association (“BCCLA”) recommends that this committee’s study clearly identify the potential harms of AI in the workplace in the following areas of impact: • Workers’ privacy interests, through increasing surveillance and data collection that may be disclosed to law enforcement; • Workers’ human rights in employment, as obscure decision making and the prospect of bias.