Google and collaborators mapped a fruit fly's brain containing 11,700 neuron types and 166,000 neurons total. Developers then used this map to create simulated fly brains that learned to play Doom and other games like Super Mario 64 and Beat Saber, with the simulated neural networks processing game inputs and controlling gameplay in real time.
Jev is a specialized AI model that produces only structured output rather than free-form text, enabling it to generate responses in parallel rather than autoregressively. This architecture delivers dramatically faster latency (70-500ms versus seconds for standard LLMs) and enables new use cases like real-time video game playing, suggesting structured output could become a fundamental computational primitive for AI systems.
A reverse-engineered model mimics Jev, TypeSafe's commercial system for selecting from multiple text options in a single pass. The repository includes implementations for Doom and chess games, with training and evaluation tools using attention-based scoring across option-context pairs.
Jev is a specialized AI model that produces only structured output instead of human language, enabling it to generate responses in parallel within a single forward pass rather than autoregressively token-by-token. This architecture makes Jev significantly faster (70-500ms vs. seconds for traditional LLMs) and opens new possibilities for AI applications, such as real-time game playing, though the author notes that similar performance could theoretically be achieved with existing LLM optimization techniques.
Rockbox and Doom have been successfully ported to run on iPod nano 3rd generation devices. The port is currently under review, with some flash models not yet supported.
Jev is a specialized AI model that produces only structured output instead of human language, enabling dramatically faster response times (70-500ms) and parallel processing in a single forward pass. This architectural difference unlocks new computational possibilities beyond traditional autoregressive LLMs, such as real-time game playing, though the author argues similar performance could be achieved through optimized structured output on existing models.
Google mapped an adult male fruit fly's brain into a 3D model containing 166,000 neurons, prompting engineers to train it to play Doom, Beat Saber, Mario 64, and Minecraft. Similar brain mapping projects like Flywire have enabled comparable applications, raising questions about future possibilities for replicating human neural structures.
A discussion on Hacker News about what users do while waiting for LLMs to complete tasks, with mentions of playing Doom, watching ads for rewards, or running multiple agent sessions.