Here are some stories and articles we followed in the last week.

A search for your next AI policy

AGORA (AI Governance and Regulatory Archive) was just recently released to help you navigate and explore the current AI policy landscape. While mostly US based, you can read about which AI policies are included in the tool and also download the full dataset.

A Canadian AI takeoff (or is it?)

The Dais released Waiting for takeoff which took a look at productivity of AI adopters in Canada 1-2 years after adoption. A keynote is that this report focused on 2020-2021 years, so pre-ChatGPT.

Here are the key findings:

Evidence of productivity gains from AI use is mixed. There is no conclusive evidence of a strong positive or negative relationship between AI adoption and short-term productivity improvement.

There was no significant relationship between the adoption of AI in this period and either Total Factor Productivity (TFP) levels or TFP growth (efficiencies in output production which do not stem from added labour or capital inputs) in the short-term.

The set of firms that adopted AI were already more productive than their peers, but the decision to adopt AI did not increase the rate at which their productivity grew.

It would be interesting to see this report updated given the current LLM AI era we are in.

The EU encryption debate continues

This article adds some commentary to the ongoing debate about message encryption occurring in the EU (and frankly across the world). In the article, the author brings up some comments and concerns about client-side scanning.

Effectively, client-side scanning would see scanning of messages occur on the device prior to the content being end-to-end encrypted for transmission. The Global Encryption Coalition addressed this issue in September of this year, discussing the limited effectiveness and increased security concerns of such techniques. You can read the response here.

These discussions are always nuanced and it will be interesting to follow this one.