Diplostack #8: AI - No Dancing in the Streets of San Francisco
Quipped the EU's Special Envoy for Digital on new digital and AI rules for doing business with 450m EU citizens ... do they strike the right balance between regulation & innovation?
In collaboration with the EU, Diplosphere brought the block’s senior digital person – Senior Envoy for Digital to the U.S. and head of the EU office in San. Francisco, Gerard de Graaf to New Zealand. Gerard spoke at Diplosphere’s recent AI conference in Wellington on 29 Aug 2023 (agenda) in front of about three hundred guests from business, government and the concerned public.
You can listen to Gerard’s on Spotify:
(All Human)
Here are Diplosphere’s notes:
There was no dancing in the streets of San Francisco - When the EU's Digital Services Act came into force last week, but EU Special Envoy says Silicon Valley respects the EU and its 450m consumers where Big Tech make big bucks, they cannot just walk away from the EU [aside - not the same in NZ!]
Product Safety Approach - This is the thinking that EU lawmakers are taking to AI. Similar to consumer guarantees and expectations with physical products, AI applications which materially affect EU citizens’ lives will have to demonstrate how and why decisions are made.
Things are moving so fast [when it comes to AI] - Nobody knows in Silicon Valley what will happen with AI past 1-2 months let alone a year.
The Silicon Valley of Regulation - This is a long-running jibe in Silicon Valley at the EU. However in recent times, the tune has changed, it is no longer a question of regulate vs. don’t regulate, but how do you regulate (AI) and when is it coming? Since the advent of Large Language Models and ChatGPT, the question has changed from why are you regulating to when will it ["the AI Act”] be ready for companies to work with? (the answer: EU body deliberations over by the end of 2023, by when companies can start to prepare using a voluntary “AI Pact” brought forth by Commissioner Breton [link]. The AI Act itself will come into force in 2026)
The Brussels Effect - The EU has global impact. As a first mover, and because AI is values-embedded (especially LLMs), the EU, which shares liberal democratic values with other states including NZ, is likely to be closely watched around the world. It seems New Zealand has adopted a “wait & see” approach to AI settings (maybe “fast follower”?). The EU contrasts its approach to AI to China as evidenced by the Commissioner’s talk, and in communiques.
Polluter pays - Just as banks and telcos pay for their own supervision, their license to operate, so it will be for AI applications which fall under the AI Act (once passed. The funding for the oversight and compliance of the affected applications of AI. The EU is assembling quite the team to look at things like algorithmic transparency, compliance to digital and AI legislation, and coordination with existing regulation.
The majority of applications using AI are risk-free. This is not the case for some - Only 10% of AI applications are likely to be caught up by the AI Act since most AI applications do not pose material risk to persons (however applications in areas such as health, mortgage decisions, recruitment, and others will have to be to demonstrate no bias and explain decisions).
Transparency key to Trust - (yet most of what Large Language Models generate is unexplainable, today …). And recent polling shows that New Zealanders are skeptical about AI, and lag in their understanding of the technology - compared to the rest of the world (Global Views on AI IPSOS, June 2023). AI has a trust issue in NZ. Whilst Diplosphere’s recent AI event polling showed 79% rated AI as high and urgent on the political agenda (n=48).
What is your P(doom)? - This is a common question now at Silicon Valley dinner parties — what is the probability of doom from AI run amok (a la Skynet/Terminator, Matrix, films etc.)?
(AI & Human collab)
Here is ChatGPT (GPT 4.0)’s Summary of the Voice-to-Text Transcript, after being chunked up into sections, to handle token limitations.
1. A year ago, Silicon Valley criticized EU's AI regulation plans; now the mood has flipped, with some talking of the existential risks posed by AI
2. EU aims to enact AI regulations by year-end and is working on an 'AI Pact' to bridge a two-year transitional gap. The focus is on product-safety and a risk-based approach, targeting the riskiest 10% of AI applications.
3. In contrast, the U.S., especially California, focuses on liability—asking who's accountable if things go wrong. Both approaches have merits, but EU's is more comprehensive and includes broad regulations like DSA and DMA.
4. Regulation is a costly, complex task with a focus on 'single-purpose AI' initially. Market surveillance and 'polluter pays' fees are in the pipeline across various sectors, including finance and healthcare.
5. Despite the rush to regulate, the EU recognizes the potential for tech to solve its own problems, especially in content moderation and cybersecurity. Silicon Valley believes advancements will benefit defense more than offense.
6. GPT-4 was used in a significant cyber attack recently, showcasing that tech can also exacerbate issues. There's a call for tech to not only solve problems it created but to also co-exist with effective policy and regulation.
7. The EU is open to learning from other regulatory models, including New Zealand, and is committed to a balanced approach that ensures AI serves society without governing it.
(All AI)
From Deciphr, an Australian startup.
The EU's Approach to Regulating AI: A Roller Coaster Ride
In the fast-paced world of technology, it's hard to keep up with the latest trends and developments. One year ago, the talk in Silicon Valley was all about the Metaverse, but now the focus has shifted to AI and social media platforms. As the EU envoy to the US, Gerard DeGraff has had a front-row seat to these changes. In this article, we will explore the EU's approach to regulating AI and the potential impact it may have on businesses and society. We will delve into the key themes discussed by DeGraff in a recent interview, providing a comprehensive analysis of the EU's stance on AI regulation.
The Roller Coaster Ride of AI Regulation
When DeGraff arrived in Silicon Valley a year ago, the discussion revolved around the Metaverse. However, the landscape quickly changed, and AI became the hot topic. DeGraff acknowledges that nobody really knows where AI is headed, but one thing is certain: it will be disruptive. The EU has taken a proactive approach to regulation, recognizing the need to balance innovation with the protection of individuals and society.
DeGraff emphasizes that the EU is not solely focused on regulation. The union is also investing heavily in technology and other areas related to AI, such as skills development and labor market impact. The EU understands that regulation alone is not enough; it must also foster innovation and ensure that AI aligns with European values. DeGraff highlights the importance of trust in AI, as a lack of trust can hinder market growth and innovation.
The EU's Regulatory Framework
The EU has been at the forefront of AI regulation, building on its previous success with the General Data Protection Regulation (GDPR). The Digital Services Act (DSA) and the Digital Markets Act (DMA) are groundbreaking pieces of legislation that aim to regulate social media platforms and ensure fair competition in the digital market. The EU's regulatory efforts extend beyond AI, with mandates on USB-C devices and other areas.
But how did the EU manage to get all member states to agree on these regulations? DeGraff explains that the EU is not obsessed with regulation for the sake of it. The union recognizes the need for comprehensive regulation to address the risks and challenges posed by AI. By involving industry experts, universities, and conducting thorough problem analyses, the EU ensures that its regulations are well-informed and effective. DeGraff also emphasizes the importance of the EU's democratic process, which includes public consultation and checks and balances.
Reception in Silicon Valley
The EU's regulatory efforts have not gone unnoticed in Silicon Valley. While some initially criticized the EU for regulating AI, the sentiment has shifted. Companies like Meta and Amazon recognize the importance of the EU market and prefer regulation from a democratic society like the EU rather than from autocratic countries. DeGraff acknowledges that there has been a constructive dialogue with industry leaders in Silicon Valley, and they understand the inevitability of regulation. The EU's regulations provide predictability and trust in the market, which can ultimately drive innovation.
Implications for Businesses and Society
The EU's AI regulations will have implications for businesses worldwide, including those in New Zealand. Compliance with EU regulations will be necessary for companies looking to access the EU market, which consists of 450 million consumers. While there will be costs associated with compliance, the EU is mindful of striking a balance between regulation and innovation. The EU's comprehensive approach to regulation ensures that risks are mitigated while maximizing the benefits of AI.
New Zealand, as a small nation, can learn from the EU's experience in regulating AI. By observing the EU's approach, New Zealand can apply best practices and tailor them to its own context. The EU's emphasis on trust, transparency, and fairness in AI can serve as a guiding principle for New Zealand's own regulatory efforts. Collaboration between the EU and New Zealand will be crucial in shaping the future of AI regulation and ensuring that it aligns with democratic values.
Conclusion and Future Outlook
The EU's journey in regulating AI has been a roller coaster ride, with the landscape constantly evolving. The EU's proactive approach to regulation reflects its commitment to protecting individuals and society while fostering innovation. The EU's regulations will have a significant impact on businesses and society, both within the EU and globally. Compliance with EU regulations will be necessary for companies looking to access the EU market, and New Zealand can learn from the EU's experience to shape its own regulatory framework.
As AI continues to advance, the EU will remain at the forefront of regulation. The EU recognizes that technology is not neutral and that AI must conform to European values. The EU's regulations provide a framework for trustworthy AI, which can drive innovation and ensure a level playing field in the global market. While the future of AI remains uncertain, the EU's commitment to regulation and collaboration with partners like New Zealand will shape the path forward.
In this rapidly changing technological landscape, the EU's approach to regulating AI serves as a model for other nations. By striking a balance between regulation and innovation, the EU aims to maximize the benefits of AI while minimizing risks. As the world grapples with the challenges and opportunities presented by AI, the EU's comprehensive and inclusive approach provides a roadmap for responsible and ethical AI development.
Event Coverage:
MPs concerned AI not well used (The Post, 5/9/23)
Is New Zealand flying blind on generative AI? (The Post, 12/9/23)
Political parties put AI regulation in the too hard basket (Business Desk, 31/8/23, behind paywall)
Can the EU's new rules really rein in the worst of Big Tech? (RNZ, 28/8/23)
Putting big tech on notice, with EU’s envoy to Silicon Valley Gerard de Graaf (Spotify: The Business of Tech podcast, 23/8/23)
Tap the Like button, and use the Share button to share across social media. And feel free to leave a comment if you are so inclined, and remember to subscribe if you haven't done so already.