The Adversarial Testing Podcast
From data curation to production monitoring — how frontier labs evaluate, red-team, and decide when to ship their most powerful models.
Sei die erste Person, die kommentiert
Melde dich jetzt an und werde Teil der The Adversarial Testing Podcast-Community!
Dann 4,99 € / Monat · Jederzeit kündbar.
5 Folgen
Electoral Hallucinations: Safeguarding UK Elections in the World of LLMs and AI Chatbots (Executive Summary)
The executive summary of Electoral Hallucinations by Jamie Hancock and Azzurra Moores, published by Demos in May 2026. The report presents new evidence from testing five AI services during the 2026 Scottish Parliament elections, finding that 34.1% of responses contained factual errors — including hallucinated candidates, incorrect voting procedures, and fabricated political scandals. It identifies a regulatory gap where AI meets elections and sets out four recommendations for the UK government ahead of 2029.
The Labour Party Is Playing With Fire Over Its Future and the Future of the Country
Tony Blair argues that Labour risks electoral irrelevance by governing from a traditional soft-left comfort zone while the world undergoes two epochal shifts. He makes the case for a Radical Centre strategy built around technological transformation, economic competitiveness, and a renegotiated relationship with Europe.
Evaluating and Testing Frontier LLMs — The Full Lifecycle
How to Train a Frontier LLM — The Full Pipeline
A technical walk-through of the entire training pipeline for a modern frontier large language model, from raw data curation through pre-training, mid-training, GRPO reasoning RL, safety alignment, and deployment monitoring.
The AI Economy Debate: What the Evidence Actually Shows
Same technology, same evidence, twentyfold gap in macro forecasts. We walk through the empirical record on AI's economic impact — adoption, worker-level RCTs, the Danish null, Acemoglu's macro arithmetic, the Anthropic Economic Index, and where the data converges.
Kommentare
0Sei die erste Person, die kommentiert
Melde dich jetzt an und werde Teil der The Adversarial Testing Podcast-Community!