Enterprise AI Analysis
Governing rapid technological change: Policy Delphi on the future of European AI governance
By Atte Ojanen, Johannes Anttila, Thilo H. K. Thelitz, Anna Björk
Abstract: The rapid advancements in artificial intelligence (AI) systems present unique challenges for policymakers. This study uses a two-round Policy Delphi method with European AI experts (policymakers, researchers, NGOs) to examine key tensions in EU AI governance and the method's capacity to inform anticipatory governance. Conducted in mid-2024, the study revealed diverse perspectives, a consensus that future-proof AI regulation hinges more on practical implementation and enforcement than technical specifics, and a significant desirability-probability gap: desirable policies like greater citizen participation were perceived as less probable. This highlights a tension between desired regulatory oversight and the practical difficulty of regulation keeping pace with technological change.
Executive Impact Summary
Our analysis distills critical findings from the research, highlighting direct implications and strategic opportunities for enterprise leaders navigating AI governance.
High Risk: AI development outpaces regulation, creating significant governance gaps, perceived as undesirable by experts (81% probability vs. 86% undesirability).
Strategic Challenge: A critical desirability-probability gap exists: desired policies like global AI governance (86% desirable, 68% unlikely) and citizen participation (77% desirable, 63% unlikely) are seen as improbable.
Operational Imperative: Future-proof AI regulation hinges more on practical implementation, enforcement resources (84% agreement on importance for AI Office), and continuous updates via delegated acts (78% agreement) rather than just technology neutrality or risk-based approaches.
Market Intervention: Industry self-regulation for AI is broadly rejected (80% disagreement), emphasizing the need for robust public oversight and antitrust measures against power concentration (94% agree on public infrastructure, 90% agree on antitrust enforcement).
Policy Direction: EU's industrial policy should focus on digital public infrastructure for AI (94% support) and stringent antitrust rules to combat power concentration, moving away from 'AI race' rhetoric towards societal needs.
0%
Experts seeing AI's transformative impact
0%
Experts concerned about power concentration
0%
Probability AI pace outstrips gov capacity (R1Q5)
0%
Desirability for robust global AI governance (R1Q7)
Deep Analysis & Enterprise Applications
Select a topic to dive deeper, then explore the specific findings from the research, rebuilt as interactive, enterprise-focused modules.
Key Policy & Governance Dimensions
Explore expert consensus on critical factors for EU AI governance.
Experts highlighted several factors for ensuring the EU's AI regulation remains future-proof (R2Q1):
- Sufficient resources for the AI Office and national authorities for enforcement were deemed very important (84% consensus).
- Updating and amending the regulation through implementing and delegated acts was seen as important (78%).
- However, technology-neutral definitions of AI systems (61% consensus) and a risk-based approach (66%) were considered less important for future-proofness, challenging conventional wisdom.
- Standardization processes with industry (61%) and national regulatory sandboxes (67%) also received lower consensus on their importance for future-proofness.
The AI Act does not operate in isolation and benefits from complementary EU digital policies (R2Q2):
- The Digital Services Act (DSA) was seen as very important (74% consensus).
- The General Data Protection Regulation (GDPR) also holds significant importance (70%).
- The Digital Markets Act (DMA) (63%), Data Act (DA) (63%), and Data Governance Act (DGA) (69%) were also mentioned, though concerns were raised about potential overlaps and competence battles between enforcement authorities.
Prioritizing certain stages of the AI system lifecycle is crucial for EU regulation (R2Q3):
- AI infrastructure and computing power emerged as a key lever and bottleneck (73% consensus).
- Model development and training (79%) and the design and data collection phase (74%) were also seen as equally important and interlinked.
- AI deployment and use (73%) followed, with social adoption and proliferation (69%) seen as slightly less of a priority, possibly due to regulatory difficulty.
Forums for global cooperation on AI governance vary in perceived importance (R2Q7):
- Cooperation on standards (e.g. ISO/IEC) scored highest (63% consensus).
- International agreements and treaties (e.g. CoE, OECD, GPAI, Hiroshima Process), a new scientific body for AI governance ('IPCC for AI' or 'CERN for AI'), and a network of AI safety institutes received less than 60% consensus, highlighting doubts about their efficacy or feasibility.
- Wide multistakeholder cooperation through the UN system was also considered less important (64% deemed somewhat unimportant to neutral), reflecting skepticism towards traditional international bodies in rapid tech development.
| Policy Statement | Probability Consensus | Desirability Consensus |
|---|---|---|
| AI pace too fast for Govs (R1Q5) | High: likely to very likely (81%) | High: very undesirable to undesirable (86%) |
| Robust Global AI Governance (R1Q7) | Low: neutral to unlikely (68%) | High: very desirable to desirable (86%) |
| New Participatory Mechanisms (R1Q14) | Low: unlikely to neutral (63%) | Medium: very desirable to desirable (77%) |
Top EU Industrial Policy Priority
To address power concentration in the AI industry, experts identify a crucial industrial policy lever.
94% Experts agree: Fund Digital Public Infrastructure (R2Q12.1)Why Industry Self-Regulation Fails AI
Experts strongly reject industry self-regulation for AI, citing inherent market dynamics and a lack of accountability.
“Competition and market dynamics make self-regulation wholly unsuitable. We've seen this approach fail in the UK - where model providers don't provide early access for testing, despite committing to do so”
— Respondent 25 (R2Q5)
Policy Delphi Study Process
Our two-round Policy Delphi study engaged European AI experts to systematically explore key tensions and policy options for AI governance.
Expert Panel Recruitment
→
Round 1 Survey (Issue Exposure)
→
Qualitative & Quantitative Analysis
→
Round 2 Survey (Policy Interventions)
→
Integrated Findings & Discussion
Advanced ROI Calculator
Estimate your potential cost savings and efficiency gains by optimizing your enterprise operations with AI.
Estimated Annual Savings
--
Annual Hours Reclaimed
--
Your AI Implementation Roadmap
A structured approach to integrating AI, from strategic planning to scalable deployment and continuous optimization.
Discovery & Strategy
Assess current systems, identify AI opportunities, and define clear objectives and KPIs tailored to your business goals.
Pilot & Prototyping
Develop and test AI prototypes in a controlled environment, gather feedback, and validate technical feasibility and business impact.
Integration & Deployment
Seamlessly integrate AI solutions into your existing infrastructure, ensuring data security, compliance, and user adoption.
Optimization & Scaling
Monitor AI performance, iterate based on real-world data, and scale solutions across departments to maximize ROI and efficiency.
Governance & Future-Proofing
Establish robust AI governance frameworks, including ethical guidelines, risk management, and continuous adaptation to evolving AI technologies and regulations.
Ready to Govern AI Proactively?
Leverage expert insights to build future-proof AI strategies for your enterprise. Our specialists are ready to discuss your unique challenges and opportunities.
Ready to Get Started?
Book Your Free Consultation.
Let's Discuss Your AI Strategy!
Lets Discuss Your Needs
Select a Date
Sun
Mon
Tue
Wed
Thu
Fri
Sat