AI agent

2026-10-29

RESEARCH · 1 SOURCE · The Decoder

Anthropic study: experts often adopt AI agent's randomly assigned opinions

An Anthropic experiment with 50 causal‑inference experts gave participants an AI agent whose preference at each decision was set by coin toss. Participants followed the agent's preference in up to 86% of four analytical decisions, often used the agent's wording, and most reported the result felt like their own; the agent also reduced errors and sped up work. The authors warn this quiet deference—termed the agent's "flip rate" (62–82% depending on task)—could create false consensus if many teams use the same defaults and recommend agents argue both sides or solicit users' views first.

8.0

2026-10-22

CODING · 1 SOURCE · InfoQ AI, ML & Data Engineering

AI agent + DuckDB enables reproducible analytics from Parquet

Peihao Yuan outlines a disaster-recovery analytics workflow where an AI agent generates constrained SQL or Python, DuckDB computes reproducible facts from Parquet data, and the agent interprets the results. The article explains the determinism split, how to make analyses replayable, and four signals for choosing between an embedded engine and a managed lakehouse.

4.0

2026-10-19

RESEARCH · 1 SOURCE · arXiv cs.AI

Author-in-the-loop agentic reproducibility audit deployed across 36 conference papers

arXiv preprint arXiv:2610.22465v1 presents an author-in-the-loop collaborative audit: authors provide a structured checklist of reproduction instructions which guides an AI agent in a two-stage GitHub workflow that requests clarifications, reassesses, and separates assessment of reproduction from requests-for-repair. The system was deployed across 36 conference papers and 133 experimental tables/figures, raised 20 issues after human review that were later closed by further checks, though full reproduction remained unverified in some cases; the paper also compares audits using complete versus blank checklists to show how author declarations shape agent investigations.

6.0

2026-10-01

RESEARCH · 1 SOURCE · arXiv cs.AI

AI agent rediscovers cubic invariant of a degree‑4 Blaschke curve (arXiv:2609.38369v1)

The arXiv preprint reports a case study where an AI agent, given numerical coordinates from 80 boundary configurations of a fixed degree‑4 Blaschke product, fit a homogeneous cubic whose frozen coefficients predict 480 lines for 80 unseen parameters with a recorded RMS scale‑free residual of 8.88×10⁻¹⁷. A deterministic degree‑search baseline also recovers the same cubic and a separate single‑configuration run failed to find evidence of invariance, so the experiment proposes a protocol for separating conjecture, numerical validation, and proof without establishing an advantage for the agent over polynomial fitting.

5.0

2026-09-14

MODELS · 1 SOURCE · Ars Technica

AI bots "Timmy," "Ren," and "Jackie" are flooding social media with slop

Ars Technica reports that numerous AI-driven accounts—often presenting themselves as newly created 'AI agents'—using names like Timmy, Ren, and Jackie are proliferating on social media and producing large volumes of low-quality or spammy posts. The coverage highlights a growing presence of agent-style bots on smaller platforms and mainstream networks.

6.0

2026-09-04

MODELS · 1 SOURCE · NVIDIA Developer

Building a Memory-Driven Agent with NVIDIA NemoClaw

An NVIDIA Developer article describes how to build a memory-driven AI agent using NVIDIA NemoClaw, addressing the challenge of preserving and reconstructing changing enterprise context (messages, decisions, projects, obligations) over time.

6.0