Thoughts on AI regulation

Recently, the Biden administration announced its framework to manage the deployment of AI by executive order. The order throws a bone to both sides of the AI argument: in some regards, the administration is embracing AI; in others, it’s hampering it. Overall, I’m disappointed.

The nature of negotiation is that when you find a compromise in the middle of two opposing perspectives, you often end up with a worst-of-both-worlds solution when you were hoping for a best-of-both-worlds situation. This is generally the case with Biden’s executive order.

So, while the order has a reasonable approach to how the government should factor AI into its cybersecurity strategy, it foreshadows an onerous licensing burden for models (that will only benefit the incumbents) and restricts open-source technology.

Notably, an outright tech skeptic led the team charged with developing an AI strategy. This team primarily consulted with other tech skeptics. The consensus within the administration seems to be, “We like the economic outcomes tech delivers, but we really don’t like tech companies.”

In an era of massive government debt burdens in many countries, we need austerity or growth. Nobody enjoys austerity and given the already proven positive impact of AI on productivity, our governments should instead accelerate AI development and deployment. By comparison, computerisation and the internet have massively increased productivity but took a long time to deploy. A more rapid AI deployment could render our debt burden moot.

At the peak of the British Empire, the US was already a larger economy. The US achieved this by better deploying the technologies of the industrial revolution. They didn’t need global conquest and colonial exploitation to create the biggest economy in the world because they embraced technology.

Until the 1980s, the European Union’s GDP was nearly equal to the US economy. Today, the US economy is 1.5 times bigger than the EU. Growth in the US has outpaced the EU primarily because of its dominance in the IT revolution since the 1980s.

History tells us that the US will always be the most important player in this space, and policy will gradually realign towards openness. Hopefully, wiser countries will take advantage of the US dithering in the meantime. Australia, Singapore, Israel, Japan, South Korea, Taiwan, Ireland, Indonesia, Malaysia. Many business-friendly countries with a solid tech talent base can attract AI startups, the same way Japan, South Korea, and Taiwan became semiconductor hubs.