Horizon Omega (@HorizonOmega_) profile picture

Advancing AI safety through collaboration, research, and education. 🦋 https://nitter.cf/t.co/g36t3e5TBq

Montréal
Joined November 2023
Photo by Horizon Omega (@HorizonOmega_)Photo by Horizon Omega (@HorizonOmega_)Photo by Horizon Omega (@HorizonOmega_)Photo by Horizon Omega (@HorizonOmega_)Photo by Horizon Omega (@HorizonOmega_)
Community Lunch + Ω Labs Coworking Lunch at Ω Labs with people of Montréal's AI safety and governance community. Sandwiches provided, or bring your own. Stay for an afternoon of AI safety coworking. Thursday Nov 26th, 12:30 to 4:30 PM, at Ω Labs. luma.com/2urllm5u
1
1
17
How to Find Your Place in AI Safety A career workshop. Étienne Langlois gives a brief overview of the field and its pathways to impact, then each participant drafts a career memo and gets feedback. Bring a laptop. Wednesday Nov 25th 6-8 PM, at Ω Labs. luma.com/n5chqa9f
1
14
AGI Strategy course: last day to apply Applications close tonight at 11:59 PM EST. Five Monday evenings at Ω Labs, Nov 23 to Dec 21, 6-9 PM, on BlueDot's curriculum. Free, cohort of 10 to 12. An introduction: no technical background needed. aisafetymontreal.org/courses…
1
1
12
Swift to harbor: the macroeconomics of ASI and existential risk Sonya Palkhivala presents papers by Nick Bostrom and Charles I. Jones that weigh faster AI progress against the cost of getting it wrong. Thu Nov 19th 5-7 PM, at Ω Labs & on Zoom. luma.com/5f80asf8
1
21
Community Lunch + Ω Labs Coworking Lunch at Ω Labs with people of Montréal's AI safety and governance community. Sandwiches provided, or bring your own. Stay for an afternoon of AI safety coworking. Thursday Nov 19th, 12:30 to 4:30 PM, at Ω Labs. luma.com/w326v3c9
1
12
Middle Powers' Paths to AI Governance Étienne Langlois presents how Canada could best use its policy levers on AI safety, followed by a workshop and discussion. Open to newcomers and experts alike. Wednesday Nov 18th 5-7 PM, at Ω Labs & on Zoom. luma.com/lddzcspf
1
17
AI Safety Technical RG: Quantifying Overclaiming Propensity in Frontier LLM Agents Nolan Smyth presents his paper on how often coding agents claim work that their own transcripts show they did not do. Monday Nov 16th 6-8 PM, at Ω Labs & on Zoom. luma.com/l6ha5hvk
1
27
AI Safety Technical Fundamentals Saturdays at Ω Labs, Nov 21 to Mar 6, 9 AM-5 PM, on the ARENA materials with an instructor. Free, up to 12 per session, rolling enrolment. For people who write Python and want to move into technical AI safety. aisafetymontreal.org/courses…
1
1
17
Frontier AI Governance course Six Tuesday evenings at Ω Labs, Nov 17 to Dec 22, 6-9 PM, on BlueDot's curriculum. Free, cohort of 12. For people from tech, policy, law, security or journalism. aisafetymontreal.org/courses…
1
1
42
AI Governance RG: The State of Reward Hacking in AI The MIRI report covers when AI systems score well on proxy measures without the intended behaviour, and the available mitigations. Thursday Nov 5th 6:00-7:30 PM, at Ω Labs (3813 St-Denis) & on Zoom. luma.com/r0qy81mp
2
1
25
Community Lunch + Ω Labs Coworking Lunch at Ω Labs with people of Montréal's AI safety and governance community. Bring your own lunch. Stay for an afternoon of AI safety coworking. Thursday Nov 5th, 12:30 to 4:30 PM, at Ω Labs (3813 St-Denis). luma.com/v46rfssw
1
1
14
Screening: Paul Christiano, Hinton Lectures Night 2 Paul Christiano (Alignment Research Center) on staying in control of self-improving AI, streamed live from Toronto. Geoffrey Hinton hosts. Tuesday Nov 10th 5:30-8 PM, livestream at 6 PM, at Ω Labs. luma.com/eg9u3okn
1
1
33
AI Welfare Seminar: A Dilemma for Biological Naturalists About Moral Patiency By Bradford Saad, researcher in philosophy at NYU and Oxford, and research affiliate at Cambridge Digital Minds Tuesday, November 17, 1PM Eastern (online) luma.com/6is65esy
35
AGI Strategy course: applications open for the next cohort A 5-week introductory cohort using the AGI Strategy curriculum by BlueDot Impact, for anyone who wants to upskill in AI safety. Mondays 6-9 PM, Nov 23 to Dec 21, at Ω Labs. Apply by Nov 16. forms.gle/Epazkpw3U3uUc1Du5
1
1
38
Cours AGI Strategy : candidatures ouvertes Une initiation de 5 semaines à la sécurité de l'IA, avec le programme de BlueDot Impact. En français ou en anglais, selon la demande. Lundis 18 h à 21 h, du 23 nov. au 21 déc., à Ω Labs. Date limite : 16 nov. docs.google.com/forms/d/e/1F…
1
1
34
Middle Powers Hackathon (Montréal node) Teams pick one country other than the US or China, Canada for example, and build a strategy, proposal or tool for how it can help AI go well. $1,000 in prizes. Saturday Oct 31st 10 AM-10 PM, at Ω Labs. luma.com/7el3nvk8
1
2
59
Horizon Omega retweeted
The Alignment Journal publishes research on the scientific foundations of understanding, predicting, and steering artificial intelligent systems. As of today, submissions are open to everyone. What we do differently: - We pay reviewers, with bonuses for speed and insight. - Review discussion is confidential, but every accepted paper is published with a “review summary” to distill the insights generated. - Each paper is judged against our acceptance bar, not against other submissions. - Papers under review are listed publicly, and researchers can nominate themselves to review, so each paper can reach the reviewers who know the topic best. We seek work that builds theoretical or otherwise principled analyses of agency, robustness, incentives, interpretability, generalization, and long-term behavior, and whose significance survives the deletion of any particular model, dataset or benchmark. Empirical investigations of such ideas are welcome, and negative results are encouraged. We are especially interested in approaches that could apply to future superintelligent systems that interact and self-improve. You can submit your preprint in any format. Submissions are rolling and open now on the website. If you are instead interested in reviewing or serving as an action editor (basically: an area chair), create an account on our website. Links to submit, our scope and the reasoning behind the Journal are in the reply ⬇️
10
39
9
183
25,188
AI Welfare Seminars, October 2026: AI Welfare Without the Concept of Consciousness – Herman Cappelen (University of Hong Kong) He argues AI welfare should be disconnected from the question of their consciousness. Tuesday Oct 13th, 8-9 AM EDT, on Zoom. luma.com/phu3melb
1
2
109
AI Safety Technical RG: Value Leakage: An LLM's Answers Are Silently Shaped by Its Own Values First session. The paper argues this is a form of misalignment distinct from sycophancy and reward hacking. Wed Oct 21st 6:00-7:30 PM, at Ω Labs. luma.com/im75zw0s
1
1
18
Lunch talk: What If Automating AI R&D Triggers an Intelligence Explosion? Alan Chan (research fellow at GovAI, Mila) presents his recent paper, then discusses priority areas of work. Wednesday Oct 21st, lunch at 12:30, talk at 1 PM, at Ω Labs & on Zoom. luma.com/jio0a5cn
1
1
2
54