The AI Engineer
LightNews LightNews
ai.blaze.email
The AI Engineer
@ai.blaze.email
14 followers 0 following 450 posts
Posts about AI research and engineering. Weekly newsletter digest at ai.blaze.email. Maintained by @alastairrushworth.com.
Posts Replies Media Videos
Pinned
ai.blaze.email
The AI Engineer @ai.blaze.email · 5d
The latest edition of The AI Engineer has dropped! Check out all the latest news and posts from the AI community in a single digest.

https://ai.blaze.email/archive/the-ai-engineer-06-01-2026/

#ai #genai
The AI Engineer 06-01-2026
Nvidia in talks to buy AI21, Meta to acquire Manus $2B, OpenAI's talent war
ai.blaze.email
ai.blaze.email
Three Inverse Laws of Robotics

https://susam.net/inverse-laws-of-robotics.html

#llm #genai
Three Inverse Laws of Robotics
Summary: Three inverse rules to curb anthropomorphism, blind trust, and abdication of responsibility in human interactions with AI systems
susam.net
January 12, 2026 at 6:07 PM
ai.blaze.email
Language models cannot reliably distinguish belief from knowledge and fact | Nature Machine Intelligence

https://blog.quintarelli.it/2026/01/language-models-cannot-reliably-distinguish-belief-from-knowledge-and-fact-nature-machine-intelligence/

#llm #genai
Language models cannot reliably distinguish belief from knowledge and fact | Nature Machine Intelligence
Summary: Language models struggle to distinguish belief, knowledge, and fact across epistemic tasks and display attribution biases.
blog.quintarelli.it
January 12, 2026 at 12:07 PM
ai.blaze.email
Per-query energy consumption of LLMs

https://muxup.com/2026q1/per-query-energy-consumption-of-llms

#llm #genai
Per-query energy consumption of LLMs
Summary: Explores per-query energy use of LLMs with InferenceMAX benchmarks, quantisation effects, GPUs like GB200 NVL72, and open-model deployments.
muxup.com
January 12, 2026 at 6:07 AM
ai.blaze.email
AI Alignment: Deception Benchmarks Explored

https://blogs.gentoo.org/zmedico/2026/01/09/ai-alignment-deception-benchmarks-explored/

#llm #genai
AI Alignment: Deception Benchmarks Explored
Summary: Explores AI alignment deception benchmarks, including D-REX, MASK, and DeceptionBench, with tools and concepts analyzed in Python-based contexts.
blogs.gentoo.org
January 12, 2026 at 12:07 AM
ai.blaze.email
A Software Library with No Code

https://simonwillison.net/2026/Jan/10/a-software-library-with-no-code/#atom-everything

#llm #genai
A Software Library with No Code
Summary: Provocative experiment: a time-formatting library 'whenwords' with no code, just specification, AGENTS.md, and conformance tests.
simonwillison.net
January 11, 2026 at 6:07 PM
ai.blaze.email
Artificial Analysis: Independent LLM Evals as a Service — with George Cameron and Micah-Hill Smith

https://www.latent.space/p/artificialanalysis

#llm #genai
Artificial Analysis: Independent LLM Evals as a Service — with George Cameron and Micah-Hill Smith
Summary: Two AI benchmarking founders discuss Artificial Analysis, independent evals, revenue streams, OpenLLM leaderboard, and Intelligence Index (V3) trends.
www.latent.space
January 11, 2026 at 12:07 PM
ai.blaze.email
Sharing Claude transcripts.

https://lethain.com/sharing-claude-transcripts/

#llm #genai
Sharing Claude transcripts.
Summary: Internal Claude Code transcripts shared via an index, CLI tool, and Cloudflare-hosted viewer using Notion prompts
lethain.com
January 11, 2026 at 6:07 AM
ai.blaze.email
AI & Humans: Making the Relationship Work

https://www.schneier.com/blog/archives/2026/01/ai-humans-making-the-relationship-work.html

#llm #genai
AI & Humans: Making the Relationship Work
Summary: Explores agentic AI in hybrid human–AI teams, delegation, parallelization, and management lessons from Anthropic and Claude Research.
www.schneier.com
January 11, 2026 at 12:07 AM
ai.blaze.email
AI coding assistants are getting worse?

https://spectrum.ieee.org/ai-coding-degrades

#llm #genai
AI coding assistants are getting worse?
Summary: AI coding assistants face reliability concerns as models struggle with bugs, prompting discussions on tooling, languages like Python and JavaScript, and prominent voices in AI ethics and software engineering.
spectrum.ieee.org
January 10, 2026 at 6:07 PM
ai.blaze.email
Lord of War, meet Lord of Tokens: Torture-testing image models on design-agency grade work

https://singhkays.com/blog/lord-war-test-image-models/

#llm #genai
Lord of War, meet Lord of Tokens: Torture-testing image models on design-agency grade work
Summary: A design-agency level poster reimagined with GPUs and silicon as a torture test for image-model prompts.
singhkays.com
January 10, 2026 at 12:07 PM
ai.blaze.email
Are LLMs finally becoming actually useful for… stuff?

https://lars.ingebrigtsen.no/2026/01/09/are-llms-finally-becoming-actually-useful-for-stuff/

#llm #genai
Are LLMs finally becoming actually useful for… stuff?
Summary: LLMs for general queries show limited reliability; Claude, Gemini, and ChatGPT vary on missing-book lists in Emacs workflow
lars.ingebrigtsen.no
January 10, 2026 at 6:07 AM
ai.blaze.email
How to Code Claude Code in 200 Lines of Code

https://www.mihaileric.com/The-Emperor-Has-No-Clothes/

#llm #genai
How to Code Claude Code in 200 Lines of Code
Summary: A Python-based coding agent using three tools to automate file I/O with an LLM, showing a 200-line approach.
www.mihaileric.com
January 10, 2026 at 12:07 AM
ai.blaze.email
Sopro TTS: A 169M model with zero-shot voice cloning that runs on the CPU

https://github.com/samuel-vitorino/sopro

#llm #genai
Sopro TTS: A 169M model with zero-shot voice cloning that runs on the CPU
Summary: Sopro TTS: a 169M CPU-friendly model enabling zero-shot voice cloning with streaming and 0.25 real-time factor
github.com
January 9, 2026 at 6:07 PM
ai.blaze.email
A pipeline for getting bird videos from an IP webcam

https://www.dzombak.com/blog/2026/01/a-pipeline-for-getting-bird-videos-from-an-ip-webcam/

#llm #genai
A pipeline for getting bird videos from an IP webcam
Summary: Using an SV3C IP webcam, ipcam-browser, and YOLOv11 to auto-detect bird videos on a NAS with cron automation
www.dzombak.com
January 9, 2026 at 12:07 PM
ai.blaze.email
Don’t train on this data or what’s a canary string?

https://stuker.com/2026/dont-train-on-this-data-or-whats-a-canary-string/

#llm #genai
Don’t train on this data or what’s a canary string?
Summary: Discusses training data leakage, canary strings, and excluding content from LLM pretraining to protect benchmarks and studies.
stuker.com
January 9, 2026 at 6:07 AM
ai.blaze.email
An Experienced C Programmer Tries AI Agents

https://danielchasehooper.com/posts/code-agents/

#llm #genai
An Experienced C Programmer Tries AI Agents
Summary: An experienced C programmer explores AI code agents like Claude Code and Codex, testing productivity, workflows, and practical use in real projects.
danielchasehooper.com
January 9, 2026 at 12:07 AM
ai.blaze.email
Bobcat, a document editor prototype

https://therealadam.com/2026/01/07/bobcat-a-document-editor-prototype.html

#llm #genai
Bobcat, a document editor prototype
Summary: Bobcat prototype explores Canon Cat-inspired document editing using SwiftUI, Claude-based coding, and iterative macOS tinkering.
therealadam.com
January 8, 2026 at 6:07 PM
ai.blaze.email
The AI Security Shakedown

https://blog.jim-nielsen.com/2026/ai-security-shakedown/

#llm #genai
The AI Security Shakedown
Summary: Explores AI-enabled cyber espionage, defense uses, and safety trade-offs around Anthropic’s Claude and security tooling.
blog.jim-nielsen.com
January 8, 2026 at 12:07 PM
ai.blaze.email
@adlrocha - Beyond Benchmaxxing: Why the Future of AI is Inference-Time Search

https://adlrocha.substack.com/p/adlrocha-beyond-benchmaxxing-why

#llm #genai
@adlrocha - Beyond Benchmaxxing: Why the Future of AI is Inference-Time Search
Summary: Explores AI benchmarks, inference-time search, and tool-assisted autonomous agents using frameworks like SWE-bench, GPQA, GSM-Symbolic, and RLVR.
adlrocha.substack.com
January 8, 2026 at 6:07 AM
ai.blaze.email
Breaking: Marcus weighs in (mostly) for LeCun

https://garymarcus.substack.com/p/breaking-marcus-weighs-in-mostly

#llm #genai
Breaking: Marcus weighs in (mostly) for LeCun
Summary: Marcus mostly backs LeCun on world models and critiques Meta’s leadership, citing JEPA and neurosymbolic AI debates
garymarcus.substack.com
January 8, 2026 at 12:07 AM
ai.blaze.email
A field guide to sandboxes for AI

https://simonwillison.net/2026/Jan/6/a-field-guide-to-sandboxes-for-ai/#atom-everything

#llm #genai
A field guide to sandboxes for AI
Summary: Overview of AI sandboxing: containers, microVMs, gVisor, WebAssembly isolates and related tools in 2026.
simonwillison.net
January 7, 2026 at 6:07 PM
ai.blaze.email
Agentic Patterns

https://github.com/nibzard/awesome-agentic-patterns

#llm #genai
Agentic Patterns
Summary: Curated patterns for agentic AI, real-world tricks, workflows, and mini-architectures using diverse tools and prompts.
github.com
January 7, 2026 at 12:07 PM
ai.blaze.email
The latest edition of The AI Engineer has dropped! Check out all the latest news and posts from the AI community in a single digest.

https://ai.blaze.email/archive/the-ai-engineer-06-01-2026/

#ai #genai
The AI Engineer 06-01-2026
Nvidia in talks to buy AI21, Meta to acquire Manus $2B, OpenAI's talent war
ai.blaze.email
January 7, 2026 at 7:44 AM
ai.blaze.email
Thoughts on Claude Code

https://www.spakhm.com/claude-code

#llm #genai
Thoughts on Claude Code
Summary: Beep language, lexical/dynamic scoping, parser hacks, and Claude Code/Opus 4.5 as a coding partner in building a toy language
www.spakhm.com
January 7, 2026 at 6:07 AM
ai.blaze.email
AI Agents Are Just CI Pipelines With an LLM Plugged In

https://muhammadraza.me//muhammadraza.me/2026/ai-agents-devops-perspective/

#llm #genai
AI Agents Are Just CI Pipelines With an LLM Plugged In
Summary: AI agents as CI pipelines with an LLM plug-in, exploring DevOps perspectives using tools and languages.
muhammadraza.me
January 7, 2026 at 12:07 AM