The Challenges of Automating AI Compliance Monitoring
This article discusses the challenges of automating the monitoring of AI regulatory compliance, despite having the intent and tooling to do so. The author highlights the gap between documenting the intent and actually implementing the code.
Why it matters
This article highlights the challenges of translating intent into action, even when the technical solution is relatively straightforward. It underscores the importance of overcoming inertia and making the initial decision to start a project.
Key Points
- 1The build-log timestamps intent but cannot write the code to implement it
- 2The author has repeatedly documented the plan to monitor ESMA regulatory feeds for MiCA-related updates, but has not yet created the first artifact
- 3Automation can overcome inertia by removing the need for human decision-making, but the AI compliance tool requires that initial decision
- 4While the automation is not running, the regulatory landscape continues to evolve, creating a gap between intent and action
Details
The article explores the pattern of repeatedly documenting the intent to build an AI compliance monitoring tool, but not actually implementing the first functional artifact. The author contrasts this with their experience running automated trading bots, where the cron job runs without requiring a decision at runtime. In the case of the compliance tool, the author notes that the complexity is not technical, but rather the initial decision to start the project. Despite the regulatory landscape continuing to evolve, with ESMA publishing new technical standards, the compliance tool remains in a state of intent rather than implementation. The author sets a target of creating the first commit with a stubbed function and test by the end of the week, as a way to overcome the inertia and make the first step towards the full implementation.
No comments yet
Be the first to comment