Today I'm publishing a new essay, Policy on the AI Exponential. AI is progressing extremely fast—much faster than the policy process was built to handle. The essay lays out where I think the technology is now, and the action needed to close the gap: https://t.co/Lh6PWae178
guys agent frameworks are so stupid, all you need is an anthropic API key and a while loop
…and a FastAPI server so that we can use the agents programmatically
…and some good API designs to enable multi user and multi agent support
…and a tool execution sandbox so that the agent tools don’t interfere with the main server process
…and a real database so that the agents are actually persisted and don’t disappear when the script finishes
…and an ORM so that we can properly scale the agent server
…and some extra middleware code since we also want to use local LLMs but they have less reliable function calling
…and some sort of file storage / embedding solution to do RAG
…and some sort of context management system to handle the long term memory problem and deal with context overflow
…
…congratulations, you just built an “agents framework”
We built a GPT-4o-powered cleaning robot.
- $250 for the robot arms
- 4 days to build
Open source is truly democratizing the field of robotics.
@KasparJanssen
The data truthers are back. Conspiracy-mongering about federal statistics is bad and harmful.
Measuring national stats is inherently difficult & subject to error. But the BLS is an independent statistical agency that produces accurate, high-quality data.
1/6 The Nature paper highlighted by @alexandr_wang raises concerns about synthetic data and model collapse. However, it overlooks the transformative potential of synthetic data in AI. Far from causing collapse, modern approaches enhance model performance and capabilities.
1- Synthetic data will help AI go the last mile. A substantial portion of the data required to train large models for the ultimate user experience and customization is not in the public domain. Businesses need to find safe methods to experiment with various foundation models