General Intuition | The frontier research lab dedicated to games—and the real world.
generalintuition.com · 358 words · saved by 2 readers
We build foundation models for environments that require deep spatial and temporal reasoning.
Machine intelligence, trained primarily on the archive of human language, inherits both its brilliance and limitations: planar abstraction without embodiment, prediction without participation. This compromise was initially pragmatic. Rather than parse the messiness of reality, models learned from its readily available, albeit reductive effigies: descriptions, depictions, and recordings. But truly intelligent machines must acquire the capacity to perceive, anticipate, and improvise within the unfolding dynamics of reality. From the relationships of intent, action, and consequence, to the…
saved by
related reading
- [1911.01547] On the Measure of Intelligencearxiv.org
- Why Tool AIs Want to Be Agent AIs · Gwern.netgwern.net
- World Models: Computing the Uncomputablenotboring.co
- The Artificial Intelligence Revolution: Part 1 - Wait But Whywaitbutwhy.com
- As Rocks May Think | Eric Jangevjang.com
- Andrej Karpathy — AGI is still a decade awaydwarkesh.com
- A Workaphile’s Apologymoalquraishi.wordpress.com
- 2023 letter | Zhengdongzhengdongwang.com
- Building machines that learn and think like people | Behavioral and Brain Sciences | Cambridge Corecambridge.org
- Artificial General Intelligence Is Already Herenoemamag.com
- Ilya Sutskever — We're moving from the age of scaling to the age of researchdwarkesh.com
- DeepMind: Generally capable agents emerge from open-ended play — LessWronglesswrong.com