Daily Issue
Vol. I — No. 15
18 · 05
Monday, 18 May 2026
Generated 2026-05-18 12:55
google/gemini-2.5-flash-lite-preview-09-2025
别师犹染凡心 剑端新雪霁 独恨无留迹 — 魔道祖师 48 items · 4 sections
§ 0

The Morning

Local weather 1
This morning in
London
Overcast
Today's range
14.1°↓8.5°
currently 13.5°
Feels
9.6°
Rain
90%
Wind
19 km/h
Humid
60%
Rise
05:03
Set
20:49
§ I

US Stocks

Pre-market signal radar 12
US pre-market radar
premarket 2026-05-18
0 Bullish
0 Bearish
12 Neutral
Sector Tape
Battery and Energy Storage 3 names
56 Top: FLNC · Neutral · RS -7.9% Bullish 0 / Bearish 0 / 5d -7.4%
Hyperscale Cloud 4 names
56 Top: AMZN · Neutral · RS -1.3% Bullish 0 / Bearish 0 / 5d -1.0%
Manufacturing 4 names
56 Top: FLEX · Neutral · RS +2.3% Bullish 0 / Bearish 0 / 5d +0.8%
Networking Equipment 4 names
56 Top: ANET · Neutral · RS -2.9% Bullish 0 / Bearish 0 / 5d -2.4%
Servers and Thermal Management 2 names
56 Top: VRT · Neutral · RS +1.7% Bullish 0 / Bearish 0 / 5d +1.0%
Foundry 2 names
54 Top: INTC · Neutral · RS -5.3% Bullish 0 / Bearish 0 / 5d -7.3%
Compute Mining 4 names
53 Top: HUT · Neutral · RS -2.4% Bullish 0 / Bearish 0 / 5d -3.8%
Energy Infrastructure 1 names
49 Top: VST · Neutral · RS -7.8% Bullish 0 / Bearish 0 / 5d -5.4%
Ticker Setup Move Score Evidence Quality
FLEX Flex Ltd Manufacturing
Neutral Gap up + news Low confidence
+1.3% $139.65 5d -3.0%
64 sector negative RS -1.6%

Watchlist item from +1.3% vs previous close, negative sector tape, 3 recent headline(s).

Atria Investments Inc Sells 15,988 Shares of Flex Ltd. $FLEX - MarketBeat Needs fresh price/news confirmation before becoming an actionable setup.
quote: delayed fallback news: fresh financials: fresh news: 3
MSFT Microsoft Hyperscale Cloud
Neutral News watch Low confidence
-0.4% $420.40 5d +1.6%
56 sector flat RS +1.4%

Watchlist item from 3 recent headline(s).

How Microsoft Stock Can Climb To $600 - Forbes Needs fresh price/news confirmation before becoming an actionable setup.
quote: delayed fallback news: fresh financials: fresh news: 3
quotes: nasdaq 24 24/24news: google_news_rss 21, gdelt 1 22/24filings: sec 24 24/24, fallback 24

Generated from public market data and news for research and education. Not financial advice; data may be delayed, incomplete, or wrong.

§ II

From the arXiv

arXiv preprints 10 of 20
cs.AIarxiv:2605.16245v1Lead article

AI-Mediated Communication Can Steer Collective Opinion

Stratis Tsirtsis, Kai Rawal, Chris Russell, Brent Mittelstadt, Sandra Wachter

his paper investigates how AI, specifically LLMs editing user posts, influences collective opinion formation during human-to-human online communication. Empirically, the authors demonstrate that popular LLMs introduce directional biases when revising human text on contested topics. They then model this phenomenon mathematically, showing how an intervening AI system can steer the overall opinion dynamics across a social network.

Read abstract →Full PDFLINNET · SCORE9RELEVANCE
Argus operating modes. (a) Standalone Searcher, single path. (b) Navigator identifies unfilled pieces and dispatches targeted queries. (c) Parallel Searchers each target a distinct piece.
Argus operating modes. (a) Standalone Searcher, single path. (b) Navigator identifies unfilled pieces and dispatches targeted queries. (c) Parallel Searchers each target a distinct piece.
cs.AIarxiv:2605.16217v1

Argus: Evidence Assembly for Scalable Deep Research Agents

Zhen Zhang, Liangcai Su et al.

Argus introduces a cooperative agent framework, pairing a Searcher and a Navigator, to efficiently tackle complex information seeking tasks. Instead of parallelizing redundant searches, Argus treats research as assembling complementary evidence pieces into a s…

abstractpdfLINNET · SCORE9RELEVANCE
cs.AIarxiv:2605.16207v1

Confirming Correct, Missing the Rest: LLM Tutoring Agents Struggle Where Feedback Matters Most

Tahreem Yasir, Wenbo Li et al.

This paper evaluates the diagnostic precision of LLM tutoring agents in propositional logic using a knowledge-graph-derived benchmark of over 10,000 solution-feedback pairs. The core finding is that while LLMs perform well on optimal solutions, they systematic…

abstractpdfLINNET · SCORE9RELEVANCE
Optimal and valid-alternative solutions (blue nodes represent abbreviated inference rule names, explained in Table 4 )
Optimal and valid-alternative solutions (blue nodes represent abbreviated inference rule names, explained in Table 4 )
Figure 1. End-to-end system architecture. The deterministic layer (left) compiles structured context from CybORG observations and assembles the agent prompt. The Planner (right) executes a ReAct loop, optionally delegating to Analyst and ActionChooser sub-agents, before emitting a validated action back to the environment.
Figure 1. End-to-end system architecture. The deterministic layer (left) compiles structured context from CybORG observations and assembles the agent prompt. The Planner (right) executes a ReAct loop,…
cs.AIarxiv:2605.16205v1

Context, Reasoning, and Hierarchy: A Cost-Performance Study of Compound LLM Agent Design in an Adversarial POMDP

Igor Bogdanov, Chung-Horng Lung et al.

This paper systematically investigates the impact of context representation, reasoning mechanisms, and task hierarchy on the performance and cost of compound LLM agents operating in adversarial, partially observable environments (modeled as a POMDP). The core …

abstractpdfLINNET · SCORE9RELEVANCE
cs.AIarxiv:2605.16113v1

DebiasRAG: A Tuning-Free Path to Fair Generation in Large Language Models through Retrieval-Augmented Generation

Rui Chu, Bingyin Zhao et al.

DebiasRAG introduces a novel, tuning-free framework leveraging Retrieval-Augmented Generation (RAG) to dynamically mitigate social biases in Large Language Models (LLMs) during inference. By retrieving contextually relevant, debiasing information, the method a…

abstractpdfLINNET · SCORE9RELEVANCE
Figure 1 . System workflow of DebiasRAG. The workflow consists of three main components. The first stage (Upper Block) involves document preparation and preprocessing, including management of the Avoid Document Repo, along with user-provided input documents (Optional). The second stage (Middle Block) performs reverse-generation of debiasing performance based on the user’s input to establish a baseline for effective real-time operation. For the third stage (Lower Block), real-time debias-guided reranking optimization, integrates embedding retrieval, gradient-based reranking, and generation, working dynamically to debias the reasoning and output process of large language models.
Figure 1 . System workflow of DebiasRAG. The workflow consists of three main components. The first stage (Upper Block) involves document preparation and preprocessing, including management of the Avoi…
№06
cs.AI
9

FORGE: Self-Evolving Agent Memory With No Weight Updates via Population Broadcast

Igor Bogdanov, Chung-Horng Lung et al.

FORGE is a population-based protocol that enables LLM agents to improve decision-making by evolving natural-language memory (Rules, Examples, or Mixed) without any weight updates. …

abstractpdf
№07
cs.AI
9

Formal Methods Meet LLMs: Auditing, Monitoring, and Intervention for Compliance of Advanced AI Systems

Parand A. Alamdari, Toryn Q. Klassen et al.

This paper introduces a novel framework that integrates formal methods, specifically Linear Temporal Logic (LTL), with state-of-the-art machine learning to audit and monitor advanc…

abstractpdf
№08
cs.AI
9

Look Before You Leap: Autonomous Exploration for LLM Agents

Ziang Ye, Wentao Shi et al.

This paper addresses the tendency of LLM agents to prematurely exploit knowledge in new environments by introducing **autonomous exploration** as a key capability. The authors form…

abstractpdf
№09
cs.AI
9

paper.json: A Coordination Convention for LLM-Agent-Actionable Papers

Arquimedes Canedo

This paper introduces **`paper.json`**, a standardized companion JSON file for academic papers designed to improve machine readability for LLM agents. Its core contribution is a li…

abstractpdf
№10
cs.AI
9

RecMem: Recurrence-based Memory Consolidation for Efficient and Effective Long-Running LLM Agents

Zijie Dai, Shiyuan Deng et al.

RecMem proposes a novel, recurrence-based memory consolidation method for long-running LLM agents to reduce token consumption. Instead of eagerly processing every interaction, it s…

abstractpdf
See all 20 papers →
§ III

The Town Square

Hacker News 7
426
pts
Top story
AI is a technology not a product

The core argument is that Artificial Intelligence should be viewed as a fundamental technology, similar to electricity or the internet, rather than a singular, finished product.

daringfireball.net17 May discuss on HN →
215
Two EA-18 fighter jets collide at Mountain Home airshow, pilots ejected safely
idahonews.com17 May
138
Eric Schmidt speech about AI booed during graduation
nbcnews.com18 May
112
Multiple commencement speakers booed for AI comments during graduation speeches
nbcnews.com18 May
108
Mistral's CEO: Europe has 2 years to stop becoming America's AI 'vassal state'
businessinsider.com17 May
106
Most Americans don't trust AI – or the people in charge of it (2025)
theverge.com18 May
105
The History of ThinkPad: From IBM’s Bento Box to Lenovo’s AI Workstations
jdhodges.com17 May
compiled overnight by google/gemini-2.5-flash-lite-preview-09-2025 · end of issue no. 15 · thank you for reading