Cherrypicked news from https://stateof.ai
Thought to Image decoding by Meta
Thoughts to image
Eureka Robotics Research
AI agent uses LLMs to automatically generate reward algorithms to train robots to accomplish complex tasks
PaLM-E: a foundation model for robotics
The model combines PaLM-540B and ViT-22B and enables as input text, images and robot states which are encoded into the same space as work token embeddings and then fed into language model to perform next token prediction.
Swift AI Drone
Champion-level Autonomous Drone Racing using Deep Reinforcement Learning (Nature, 2023)
Various models made significant advancements in the quality of controllable music
Demo by Meta:
Google's Med PaLM 2 language model
Google's Med PaLM 2 is an expert according to the USMLE (US Medical Licensing Examination),
Mathematical Theorem Proving via AI
AlphaTensor - Discovering faster matrix multiplication algorithms with reinforcement learning
by treating algorithm discovery process as a single-player game.
Predictions for the next 12 months from stateof.ai
- A Hollywood-grade production makes use of generative AI for visual effects.
- A generative AI media company is investigated for its misuse during in the 2024 US election circuit.
- Self-improving AI agents crush SOTA in a complex environment (e.g. AAA game, tool use, science).
- Tech IPO markets unthaw and we see at least one major listing for an AI-focused company (e.g. Databricks).
- The GenAI scaling craze sees a group spend >$1B to train a single large-scale model.
- The US’s FTC or UK’s CMA investigate the Microsoft/OpenAI deal on competition grounds.
- We see limited progress on global AI governance beyond high-level voluntary commitments.
- Financial institutions launch GPU debt funds to replace VC equity dollars for compute funding.
- An AI-generated song breaks into the Billboard Hot 100 Top 10 or the Spotify Top Hits 2024
- As inference workloads and costs grow significantly, a large AI company (e.g. OpenAI) acquires an inference-focused AI chip company.