2025 Air Street Capital The State of AI Report 2025

Slide locked Sign in to view

Slide schematic 5/6

callout 0/1 image 1/1 list 2/2 paragraph 1/1 title 1/1

1 without position callout 1

callout

Components 6

callout infographic bullet bullet paragraph headline

Tools 1

Deductive Reasoning slide · 70%

paragraph/paragraph: LLMs can perform "inductive out-of-context reasoning" (OOCR) - inferring hidden information scattered across training documents and applying it to new tasks.

Metrics 0

∅

No metrics

No metric insight is anchored to this slide.

Frameworks 1

inductive-reasoning 90%

Mentions 'inductive out-of-context reasoning' (OOCR) as the core mechanism.

Find anything

Discusses 'City 50337' experiment and Synthetic Document Fine-tuning (SDF) as a method for auditing LLM alignment.