Helen Jackson

Physicist, Machine learning. public speaker

[Helen Jackson @netdove]

I just published Aligning with Ethical AI Mandates Amid a Global Rift https://t.co/n0GjYAoaGE

[Helen Jackson @netdove]

Having Birthday dinner with Monks for Peace

[Anthropic @AnthropicAI]

AI can make work faster, but a fear is that relying on it may make it harder to learn new skills on the job.

We ran an experiment with software engineers to learn more. Coding with AI led to a decrease in mastery—but this depended on how people used it.
https://t.co/lbxgP11I4I

[SAROJ PARAJULI @Parajulisaroj16]

📌📚In this comprehensive guide, we will provide an overview of Bayesian modeling and computation in Python, including key concepts and popular libraries. https://t.co/Vy0mWvvwg7
#DataScience #pythonprogramming #DataScientist
#machinelearning #statisticalanalysis #dataviz

[Oliver Prompts @oliviscusAI]

Microsoft killed the GPU mafia 🤯

They finally open-sourced their 1-bit LLM inference framework called bitnet.cpp. It lets you run 100B parameter models on your local CPU without GPUs.

- 6.17x faster inference
- 82.2% less energy on CPUs

100% Open Source.

[Helen Jackson @netdove]

@heygurisingh Claude

[Helen Jackson @netdove]

@ElonMusk_LordX

[Helen Jackson @netdove]

@Rhiachnash Yes

[Helen Jackson @netdove]

@elonnewsupdatex A

[Wiz @wiz_io]

Thousands of MCP servers are already live, but most security teams don’t have a clear strategy yet. Get this guide and learn:
- Key risks with local and remote MCP servers
- Real-world threats like prompt injection and supply chain compromise
- Steps for safely using MCP tools

[Helen Jackson @netdove]

@FanElonMuskrut 500 K

[Helen Jackson @netdove]

@elonnews40 Mona Lisa

[Helen Jackson @netdove]

@ElonMusk_LordX To give your utmost to make the world better and advance civilization with the talents and gifts you have.

[Shalini Goyal @goyalshaliniuk]

RAG is not just one technique, it is an entire ecosystem of intelligence.

From context-aware assistants to domain-specific systems, here are 16 types of RAG models shaping the next wave of AI innovation -

1. Standard RAG
The foundation of all RAG systems - combines retrieval and generation for question answering and knowledge synthesis.

2. Agentic RAG
Empowers AI agents to retrieve and act autonomously, perfect for assistants that need dynamic, tool-based reasoning.

3. Graph RAG
Uses knowledge graphs for relational reasoning - ideal for expert systems in law, medicine, and semantic search.

4. Modular RAG
Breaks retrieval, reasoning, and generation into independent components - enabling collaborative, scalable AI workflows.

5. Memory-Augmented RAG
Adds persistent external memory for context retention, powering long-term chatbots and personalized experiences.

6. Multi-Modal RAG
Processes text, images, and audio together - perfect for video summarization, captioning, and multi-modal AI tools.

7. Federated RAG
Enables privacy-preserving retrieval from decentralized sources, used in healthcare and secure enterprise systems.

8. Streaming RAG
Performs real-time retrieval and generation, ideal for financial dashboards, live feeds, and social media monitoring.

9. ODQA RAG (Open-Domain QA)
Handles large, diverse datasets - ideal for search engines and intelligent virtual assistants.

10. Contextual Retrieval RAG
Maintains session-level awareness, great for conversational AI and customer support chatbots.

11. Knowledge-Enhanced RAG
Integrates structured domain data, useful for legal, educational, and professional knowledge applications.

12. Domain-Specific RAG
Custom-tailored for specific industries - like finance, healthcare, or legal analytics.

13. Hybrid RAG
Combines multiple retrieval approaches, bridging structured and unstructured data for high precision.

14. Self-RAG
Introduces self-reflection to refine its own answers, enabling AI models to fact-check and improve reasoning autonomously.

15. HyDE RAG (Hypothetical Document Embeddings)
Generates hypothetical documents to guide retrieval, excellent for complex or niche query contexts.

16. Recursive / Multi-Step RAG
Performs multiple retrieval-generation loops, enabling advanced problem-solving and reasoning chains.

From simple retrievals to self-improving AI reasoning loops, RAG is evolving fast.
Which type do you think will dominate enterprise AI systems in 2026?

[Helen Jackson @netdove]

@swapnakpanda Want a copy

[Helen Jackson @netdove]

Demystifying the science behind fission and fusion https://t.co/zMT66U1b2n

[Elias Bareinboim @eliasbareinboim]

One question I’ve received a few times, and would like to clarify about this work (https://t.co/6s4n0490ff), is: why do we need identification and the ctf-calculus? Isn’t the do-calculus enough?

The answer to the first question is that identification is essential: estimating a quantity that is not identifiable provides no meaningful answer to the question at hand. In other words, it’s simply misleading. The metaphor I’ve been using recently comes from linear algebra. If we have a system with k equations and more than k unknowns, there are infinitely many solutions. It feels odd — even irresponsible — to boldly assert that one of them should be preferred over the others. In causality, this means that if multiple causal effects are compatible with the data and assumptions, arbitrarily picking one is nonsensical. In optimization, it would be like searching for a solution and, upon finding a local optimum, declaring it the global solution. This very issue has motivated many of the field’s foundational developments, starting with symbolic methods such as the back-door and do-calculus.

As for the second question: why not just use the do-calculus? This touches the core of a fundamental issue related to expressiveness. The do-calculus was designed for layer 2 (interventional layer) of the hierarchy (refresher: https://t.co/MnlAuEh1dP). But if your query belongs to layer 3, say, the effect of treatment on the treated, or a direct or indirect effect, there’s no way to symbolically express it using only the do-operator and the language of layer 2.

Pearl’s structural approach allows us to express such counterfactual queries, but a symbolic machinery akin to the do-calculus was still missing. This paper represents a breakthrough and completes the picture begun by Judea in his seminal 1995 Biometrika paper: it introduces the ctf-calculus and proves that it is complete for various identification tasks involving both observational and experimental data. We should be happy now -- layer 3 finally receives the same VIP treatment as layer 2, which it has long and duly deserved!

@yudapearl @analisereal @soboleffspaces @VC31415 @GaryMarcus

[Matt Dancho (Business Science) @mdancho84]

🚨 BREAKING: IBM launches a free Python library that converts ANY document to data

Introducing Docling. Here's what you need to know: 🧵

[BensenHsu @BensenHsu]

The paper behind it:

Title: Neural Tangent Kernel:
Convergence and Generalization in Neural Networks

Scientists have been trying to understand how artificial neural networks, which are like simplified brains made of math, learn and get good at tasks. When these networks are very wide, they act like something called Gaussian processes at the very beginning. This paper looks at what happens when these wide networks are trained using a method called gradient descent.