New issues of JEAHIL, JCHLA (reviews of Lens, DynaMed, OpenAlex etc)
New Aaron Tay posts:
The Blank Box Problem: Why It’s Harder Than Ever to Know What to Type Into an AI Search Bar
Deep Research, Shallow Agency: What Academic Deep Research Can and Can’t Do
Google Scholar Labs (ask it a natural language question and it provides a set of citations with each having a summary of the answer underneath it)
Google Scholar blog (highlights and comments now available in Scholar PDF Reader)
Interesting read – Technically Accurate, Medically Fatal : The AI Error We Caught in Real-Time
Very extensive wiki on AI compiled by a medical librarian (Dean Guistini). Scroll down to see the most visited topics, review on individual tools
Conducting Systematic Reviews in a Day: Enter Artificial Intelligence (one of the authors is from the Centre for Journalology so must be good)
“We recently introduced otto-SR (Otto Science Institute), a generative AI system for automated screening and data extraction that incorporates advanced prompting strategies and agentic LLM workflows. Data from currently unpublished studies involving benchmarking against dual human reviewers suggest that otto-SR achieved superior performance in both screening (otto-SR: 96.7% sensitivity, 97.9% specificity; human: 81.7% sensitivity, 98.1% specificity) and data extraction (otto-SR: 93.1% accuracy; human: 79.7% accuracy) tasks. Most notably, otto-SR reproduced and updated an entire issue of the Cochrane library (12 SRs) in under 2 days*, highlighting the potential for automation to accelerate evidence synthesis and to provide decisionmakers with timely information. Across these 12 SRs, otto-SR included nearly twice as many eligible studies as the original Cochrane authors (114 versus 64 studies)”
* “Using otto-SR, we reproduced and updated an entire issue of Cochrane reviews (n=12) in two days, representing approximately 12 work-years of traditional systematic review work“. From the preprint describing Otto
Queryome: Orchestrating Retrieval, Reasoning, and Synthesis across Biomedical Literature
“More recently, the concept of agentic RAG has gained traction, promising more sophisticated “deep research” capabilities. Systems developed by industry leaders such as OpenAI [16], Perplexity AI [17], and Google [18] have demonstrated the ability to decompose complex questions, perform iterative searches, and synthesize more comprehensive reports. Yet, these general-purpose agentic systems are not specifically tailored for the biomedical domain … to bridge this gap, we introduce Queryome, a multi-agent deep research system designed specifically for end-to-end biomedical literature analysis. Queryome orchestrates a hierarchy of collaborating AI agents that perform iterative, multi-faceted searches against a curated, comprehensive search engine covering the entirety of PubMed [1]. Crucially, the system is engineered to reason over abstract text of every retrieved article, ensuring that its final synthesis is deeply grounded in the available evidence”
Available as an app for Windows & MacOS
Instats videos – quite a few are free (using Filters (top right) > Sort by > Free. Many are quite technical but there are a number on research, using statistical tools (R, Python etc). An upcoming one is AI Tools for Research 2.0 (requires free registration, can be watched later)
Previously mentioned the many useful ebooks available via Open Educational Resources (OER) but difficulty in keeping track on new ebooks available. You can sign up to receive updates here
Spotted in the Fin Review re terrible corporate jargon – being “Promoted Outwards” is “not about job cuts but giving employees the opportunity to embrace new challenges outside the organisation”