Scaling to Long-Horizons: Algorithms, Environments, Compute
Ross Taylor, Chengxi Taylor
- When
- Tuesday, June 302:25 PM – 2:45 PM · 20 min
- Where
- Track 9San Francisco, CA · imported from ai.engineer's public schedule feed
About this session
What does it take to scale language models to year long tasks? In this talk we'll cover the algorithm, environment and compute considerations for scaling language models to long horizons. We'll cover the latest reinforcement learning approaches, how to build hard, high-fidelity long-horizon environments, and how to build scalable infrastructure for these tasks.
Speakers (2)
CEO, General Reasoning
CEO at General Reasoning Inc building long-horizon AI systems. Previously reasoning lead at Meta AI, Llama 2, Llama 3, Galactica, and founder of Papers with Code (acquired by Meta).
Co-founder & President, General Reasoning Inc.
Co-founder & President at General Reasoning Inc. Building long-horizon AI systems, and evals research lead working with leading frontier labs. Previously CEO and Chief Engineer at Satori, CEO of MyMiniFactory.
More in Data Quality
- Data Quality is the Compute MultiplierTuesday, June 30 · 10:45 AM – 11:05 AM · Track 9
- The Messy Reality of Scale: Synthetic Data and Pre-Training at PoolsideTuesday, June 30 · 11:10 AM – 11:30 AM · Track 9
- Rethinking Environments for Long Horizon WorkTuesday, June 30 · 11:40 AM – 12:00 PM · Track 9
- The Base Model is DeadTuesday, June 30 · 1:30 PM – 1:50 PM · Track 9
- Ending AI SlopTuesday, June 30 · 1:55 PM – 2:15 PM · Track 9
For developers: this programme is open data — JSON, iCal, schedule XML and an MCP endpoint.Show endpointsHide
- JSONEvery published session and speaker, in one request./aie-worldsfair-2026-import/feed.json
- iCalSubscribe in Google, Apple or Outlook Calendar./aie-worldsfair-2026-import/feed.ics
- Schedule XMLfrab / pentabarf — the format conference apps import./aie-worldsfair-2026-import/feed.xml
- MCP + RESTPoint Claude at the programme. OpenAPI 3.1 included./agents
No key, no signup, CORS open. Everything here is generated from the same data the organisers edit.