The price to pay for AI: How are companies handling consumer info?

Key Concepts

• Algorithmic Models: Computer programs designed to process data, identify patterns, and make predictions or decisions. In this context, they are used by digital companies to analyze user information.
• Monetization: The process of generating revenue from a product or service. Digital companies monetize their algorithmic models by using them to personalize experiences, target ads, or sell access to insights.
• Terms of Service Updates: Changes to the legal agreements that users must accept to use a digital service. These updates often contain clauses that grant companies broader rights to user data.
• Privacy Revolution: The ongoing effort by individuals and regulators to regain control over personal data and limit its collection and use by digital companies.
• Under-enforcement: The lack of robust legal or regulatory oversight to prevent companies from engaging in practices that may harm consumers or violate privacy.
• Abuse of Dominance: A situation where a company with significant market power uses its position to unfairly disadvantage competitors or exploit consumers.
• Competition Bureau: A Canadian government agency responsible for ensuring fair competition in the marketplace.

The Hidden Cost of Digital Services: Fuelling AI with User Data

This discussion with Vass Bednar, Managing Director of the Canadian Shield Institute for Public Policy, highlights a critical and often overlooked aspect of using popular digital services: users are inadvertently fuelling the AI products of these companies, which are then sold back to them. Bednar argues that this is a "somewhat dramatic" reality, driven by subtle "terms of service updates" that allow digital platforms to collect and utilize user information to train their algorithmic models.

The Evolving Bargain: From Privacy to Data Extraction

The current situation is presented as a departure from the older, more transparent bargain where the "price of participation online was just privacy writ large." Previously, users understood that using "free" services meant receiving tailored advertisements. However, the landscape has shifted. Bednar explains that the "privacy revolution" has not been effective in preventing a new form of data extraction.

Google as a Leading Example of Data Exploitation

Google is cited as a prime example of this trend. Bednar points out that the ability for publishers and creators to be indexed and appear in Google searches is now "tied to or dependent on them using your information for their algorithmic system." This creates a dilemma for content creators who need visibility but are compelled to allow their data to be used. The transcript notes that other jurisdictions have seen class-action lawsuits or antitrust actions against Google for such practices, a conversation that is notably absent in Canada.

Market Corrections and the Need for Policy

While some companies, like Zoom and SoundCloud, have "walked things back" in response to consumer pushback, Bednar emphasizes that this is not a reliable solution. He states that "we have to take them at their word because we're also under-enforcing." This underscores the need for proactive policy intervention.

The Challenge of Reversing the Trend

The question of whether the "cat is out of the bag already" is posed, with Bednar acknowledging the difficulty of reversing the current trajectory. He observes a sense of defeatism among users who dismiss the issue as "just a little information" or accept it as "the price." Bednar argues that a "different bargain is possible" and that the intangible nature of the online economy makes it harder for individuals to recognize what is being extracted. He draws an analogy to physical transactions, suggesting that if something were being taken from users simply for entering a store, the reaction would be more pronounced.

LinkedIn's Data Retention Practices

LinkedIn is presented as another case study. Even if users "opt out" of AI features or turn them off, Bednar explains that "the past years... of your information, not just information that you type, but jobs you look at, people you creep, what times you go on – like all this information stays, of course, in their algorithmic models." Companies are reserving the right to use this historical data for their monetized algorithmic systems.

The Path Forward: Policy Leadership and Collective Action

When asked about individual protection, Bednar dismisses the idea of individuals monetizing their own data as impractical. He stresses the necessity of "policy leadership" and views this as an opportune moment for Canada to establish standards through privacy legislation and competition law. He suggests that Canadian publishers could pursue class-action suits or private cases against companies like Google for "abuse of dominance." Bednar's role, as he sees it, is to "spark a little bit more conversation" through "storytelling and diagnosing of the problem."

The Current Reality for Individuals

In conclusion, Bednar states that "right now we are along for the ride." The current bargain for using most digital systems involves agreeing that companies can collect and use both volunteered and inferred user information. The conversation ends with the acknowledgment that time is running out, but the desire to continue this critical discussion remains.