Conference Presentation, Panel, Fireside Chat
'Enterprise AI Safeguards Defining Safety and Enhancing Trust, Governance in the Age of GenAI' |
- France's national AI strategy, initiated in 2018, will maintain implementation coherence across ministries, supported by subsidies and open calls to sustain the startup ecosystem.
- An additional 40 million euros will be invested to extend national supercomputer infrastructure while efforts focus on attracting government talent to establish ecosystem metrics.
- Responsible AI is projected to be integrated into business models as a primary opportunity rather than a constraint, coinciding with the expectation that an open source ecosystem will become the foundational layer of the future AI landscape.
- The French government plans to launch an open call for "communes numériques" to build common databases for generative AI, aiming to resolve trust barriers which are identified as a major blockage to adoption in France and beyond.
- Progress in safety technologies, including watermarking, bias mitigation, and automated red teaming via synthetic data, is anticipated to continue rapidly alongside new technical challenges that will likely emerge as current issues are solved.
- The rapid pace of the AI field is expected to persist for approximately two years, necessitating dynamic governance that involves builders rather than relying on static, long-term regulations.
- Microsoft will continue releasing tools like the Parti platform and sharing cutting-edge research as open source packages to accelerate the responsible AI journey for customers and partners.
- Business investment in the EU is projected to decrease by up to 48% over a three-year period if regulatory clarity is not provided, with 31% of businesses citing this lack of clarity as a primary reason for non-adoption.
- Risk-based, use-case specific regulation is deemed critical for startups lacking resources to navigate intersecting rules, while the EU AI Act is expected to provide a safe environment by defining implementable metrics.
- A multilateral agreement is under consideration at the international level to facilitate training notifications while supporting open source models.
- Singapore will host the United Nations AI Forum in May to address regulatory topics, with local frameworks developed based on desired citizen experiences and a regulatory stance acknowledging that hallucinations cannot currently be eliminated.
- Singaporean regulators will advise citizens against placing their entire knowledge reliance on technology, emphasizing that decision-makers still face difficulties in governing new issues without direct involvement from AI builders.