Ban, Build, Brush Off: 3 Strategies Countries Are Taking on AI

Ask the AI safety crowd about the problems with "alignment" and one of the things that they might ask you is "aligned with whom?" Given that countries around the world can't together on much less complex issues that superintelligent AGI, how much hope do we have to align around an alignment in this circumstance?

Well, while that particular convo may still feel fairly far off, it is definitely the case that nations are starting to have their first reactions to this recent AI explosion. As you might expect, they're fairly different.

Ban

Of course Europe is jumping out onto the ban-wagon. For a place that has professed for years to want to be a leader in tech, they sure do have a habit of creating rules and regulations that keep tech facing firmly away. Italy took the lead with a temporary ban of ChatGPT announced in March. It argued that ChatGPT wasn't compliant around privacy with GDPR and also lacked age controls. As of today, Germany has also launched an inquiry into OpenAI privacy practices. Many speculate these issues are one of the big motivators of the new ChatGPT incognito mode which doesn't record chat history.

Build

No longer in the EU, the UK is flexing it's freedom to do whatever the hell it wants regardless of Brussels. When it comes to AI, that looks like a £100 million Foundation Model Taskforce that is meant to develop a sovereign British AI that makes the UK 'globally competitive' in this area. There are some questions about to what extent the Brits are committing to actually building an LLM competitor vs just working on integrations and AI opportunities, but in either case, they're going building.

Brush Off

.....and then there's the US. Our approach to AI is classic: assume we're above it and just spent all our time trying to mess up the crypto industry instead. In all seriousness, it seems like Senate Majority Leader Chuck Schumer has been preparing legislation. On April 13, Axios reported that Congress is considering AI guardrails that include:

1. The identification of who trained the algorithm and who its intended audience is.

2. The disclosure of its data source.

3. An explanation for how it arrives at its responses.

4. Transparent and strong ethical boundaries.

Number 3 might be a challenge, given that we kinda don't know. Anyway, if President Biden's Twitter, expect more on this soon.

Other things:

• Call Annie is real life "Her"

• BabyAGI is 1-month old

• Palantir war AI

• The Ur-Speculation that is Apple AI

This meme nails it

What We Talked About Today:

Thanks for reading! If you want more AI Breakdown:

The AI Breakdown podcast - https://pod.link/1680633614

The AI Breakdown YouTube - https://www.youtube.com/@theaibreakdown

Signing off from the future’s past - NLW