The A.I. Act is an attempt to address the risks technology poses to jobs, misinformation, bias and national security. Adam Satariano, European technology correspondent for The Times, has been reporting on regulators' efforts to draw limits around A.I. He spoke to DealBook about the challenges of regulating a rapidly developing technology, how different countries have approached the challenge, and whether it is possible to create effective safeguards for a borderless technology with vast applications.
What are the different schools of thought when it comes to regulating A.I., and what are the merits of each approach? How long do we have?
The European Union has adopted a "risk-based" approach where they define different uses of A.I. that can pose the greatest potential harm to individuals and society — think of an A.I. used to make hiring decisions or operate critical infrastructure such as energy and water. These tools face more oversight and scrutiny. Some critics say the policy falls short because it is overly prescriptive. If something isn't listed as "high risk," then it isn't covered. The European Union's approach leaves many potential gaps that policymakers have tried to fill. For example, A.I. systems More powerful ones made by OpenAI, Google and others will be able to do many different things beyond simply powering a chatbot.
There has been a very fierce debate about how to regulate this underlying technology. How would you describe the significant differences in the way the US, EU, UK and China are approaching regulation? And what are the prospects for collaboration, given events like the recent A.I. security summit? in the UK, but also the apparent fears each country has about what the other is doing?
The A.I. shows the wider differences between the US, the European Union and China in terms of digital policy. The US is much more market-oriented and practical. America dominates the digital economy, and policymakers are reluctant to create rules that threaten that leadership, especially for a technology as potentially significant as A.I. President Biden signed an executive order imposing some limitations on the use of A.I., especially as it relates to national security and deepfakes. The European Union, a more regulated economy, is being much more prescriptive about rules for A.I., while China, with its state-controlled economy, is imposing its own set of controls with things like algorithm registrations and chatbot censorship. . The UK, Japan and many other countries are taking a more practical, wait-and-see approach.
Countries like Saudi Arabia and the United Arab Emirates are investing money in developing A.I. What are their biggest concerns?
The future benefits and risks of A.I. are not fully known - to the people who create the technology or the policymakers. This makes it difficult to legislate. Therefore, a lot of work is being done to analyze the direction of technology development and establish safeguards, whether to protect critical infrastructure, prevent discrimination and bias, or prevent the development of killer robots.
How effective can A.I. regulation be?
Technology appears to be advancing much faster than regulators can devise and pass rules to control it. This is probably the quickest response I've seen policymakers around the world give to a new technology. But it has not yet resulted in many concrete policies. Technology is advancing so quickly that it is outpacing policymakers' ability to create rules. Geopolitical disputes and economic competition also increase the difficulty of international cooperation, which most believe is essential for rules to be effective.
