Curated Digest: Announcing ILIAD 2026
Coverage of lessw-blog
lessw-blog announces the third iteration of ILIAD, a mathematically focused unconference dedicated to building the scientific foundations of AI alignment.
The Hook
In a recent post, lessw-blog announced the upcoming ILIAD 2026 conference, scheduled for August 3-7, 2026, at Lighthaven in Berkeley, California. This marks the third iteration of the gathering, which originally launched in the summer of 2024. The announcement signals a continued, organized effort to consolidate the brightest minds working on the theoretical underpinnings of artificial intelligence safety.
The Context
As artificial intelligence systems scale in both capability and societal integration, the challenge of AI alignment-ensuring these systems operate safely, predictably, and in accordance with human intentions-has transitioned from a niche philosophical concern to an urgent technical necessity. Historically, much of machine learning has relied on empirical trial and error. However, as models approach advanced capabilities, empirical tinkering is widely considered insufficient. Developing robust, mathematically sound frameworks for alignment is critical for mitigating future existential and societal risks. The broader research community is increasingly recognizing that rigorous scientific foundations are required to guarantee safety at scale. This shift toward formal verification, mathematical guarantees, and foundational theory is essential for the next generation of AI development, directly impacting future AI regulation, safety standards, and enterprise adoption. Without these mathematical guarantees, deploying autonomous systems in high-stakes environments remains a profound vulnerability.
The Gist
lessw-blog's announcement details a specialized gathering specifically designed to tackle these theoretical and mathematical challenges head-on. Rather than a traditional academic symposium with rigid presentation tracks and passive audiences, ILIAD 2026 will employ an unconference format. This participant-led structure is intended to foster collaborative, emergent research directions among the 150 to 200 expected attendees. By removing the strict schedules of standard conferences, the event empowers researchers to form dynamic working groups, dive deeply into spontaneous technical debates, and address bottlenecks in alignment theory in real-time.
The focus remains strictly on the mathematical and scientific foundations of AI alignment, ensuring a high signal-to-noise ratio for theoretical researchers. Confirmed attendees already include notable figures in the field such as Richard Ngo, Abram Demski, and Adam Goldstein, indicating strong community buy-in. Furthermore, the organizers are prioritizing accessibility and talent acquisition by offering free tickets and needs-based financial support for travel and accommodations, ensuring that financial barriers do not prevent top-tier researchers from contributing.
Conclusion
For researchers, theoreticians, and practitioners focused on the long-term trajectory of artificial intelligence, this event represents a significant node of collaboration. The push for a mathematically grounded approach to AI safety is one of the most important scientific endeavors of our time, and gatherings like ILIAD are crucial for accelerating this work by breaking down institutional silos and fostering direct, high-bandwidth communication among experts. To learn more about the event structure, the specific theoretical focus areas, or how to apply for attendance, read the full post on lessw-blog.
Key Takeaways
- ILIAD 2026 is scheduled for August 3-7, 2026, at Lighthaven in Berkeley, marking the third iteration of the AI alignment conference.
- The event maintains a strict focus on the mathematical and scientific foundations of AI alignment, moving beyond empirical approaches.
- An unconference format will be utilized to encourage participant-led programming and emergent research collaboration among the 150 to 200 attendees.
- Notable figures such as Richard Ngo, Abram Demski, and Adam Goldstein are confirmed to attend.
- To ensure accessibility for top researchers, tickets are free and needs-based financial support is available for travel and accommodations.