Tech & Society
Songs Made to Order: Text-to-music services evolve with Udio and Suno’s customized song creations.
A new breed of audio generator produces synthetic performances of songs in a variety of popular styles.
Latest
Machine Learning Research
Tuning LLMs for Better RAG: Meta’s RA-DIT boosts language model output by optimizing text retrieval
Retrieval-augmented generation (RAG) enables large language models to generate better output by retrieving documents that are relevant to a user’s prompt. Fine-tuning further improves RAG performance.
Tech & Society
GPT Store Shows Lax Moderation: A report exposes policy violations in OpenAI’s GPT Store.
OpenAI has been moderating its GPT Store with a very light touch. In a survey of the GPT Store’s offerings, TechCrunch found numerous examples of custom ChatGPT instances that appear to violate the store’s own policies.
Tech & Society
Hallucination Creates Security Holes: Researcher exposes risks in AI-generated code
Language models can generate code that erroneously points to software packages, creating vulnerabilities that attackers can exploit.
Tech & Society
Custom Agents, Little Coding: All about Google’s Vertex AI Agent Builder
Google is empowering developers to build autonomous agents using little or no custom code.
Letters
Agentic Design Patterns Part 5, Multi-Agent Collaboration: Prompting an LLM to play different roles for different parts of a complex task summons a team of AI agents that can do the job more effectively.
Multi-agent collaboration is the last of the four key AI agentic design patterns that I’ve described in recent letters.
The Batch Newsletter
AI Agents With Low/No Code, Hallucinations Create Security Holes, Tuning for RAG Performance, GPT Store’s Lax Moderation
The Batch AI News and Insights: Multi-agent collaboration is the last of the four key AI agentic design patterns that I’ve described in recent letters. Given a complex task like writing software, a multi-agent approach would break...
Data Points
Big Updates for GPT-4 Turbo, Gemini 1.5, Mixtral, and More: Plus, AI Helps Rebuild Lost Memories
This week's top AI news and research stories featured all about Google's Vertex AI Agent Builder, security holes in generated code, a series of policy violations in the GPT Store, and RA-DIT, a fine-tuning procedure that trains an LLM and retrieval model together to improve the...
Tech & Society
What Users Do With Generative AI: Studies from Filtered and Pew uncover trends in generative AI usage
Generative AI is being used mostly to generate ideas. The tech consultancy Filtered studied the most common uses for generative AI. While most gen AI users produced text, the study surprisingly found that users were slightly more likely to generate videos than images.
The Batch Newsletter
Autonomous Coding Agents, Instability at Stability AI, Mamba Mania, What Users Do With GenAI
The Batch AI News and Insights: Planning is a key agentic AI design pattern in which we use a large language model (LLM) to autonomously decide on what sequence of steps to execute to accomplish a larger task.
Data Points
Serverless GPU Inference: Plus, Fine-tuning and custom models by OpenAI
This week's top AI news and research stories featured the proliferation of coding agents, a study showing the most common uses for generative AI, the instability of Stability AI, and a transformer alternative called Mamba. But first:
Letters
Agentic Design Patterns Part 4, Planning: Large language models can drive powerful agents to execute complex tasks if you ask them to plan the steps before they act.
Planning is a key agentic AI design pattern in which we use a large language model (LLM) to autonomously decide on what sequence of steps to execute to accomplish a larger task.