Thinking in the open.
Notes on how we test aviation AI: reducing hallucinations, understanding concepts, and planning flights accurately.
Coming Soon — AeroEmbed-1.5
AeroEmbed-1.5 is the next step for AeroAI's compact aviation model, keeping the 0.8B footprint while improving reliability, reasoning, and tool use.
Introducing AeroEmbed-1
A closer look at AeroAI's 0.8B-parameter aviation model, its prompt format, tool calling, and the limits that matter for safety.
Why We Use the Gemma 4 Family
Strong capability at practical model sizes makes Gemma 4 a useful foundation for AeroAI's aviation model work.
AvBench 1.0 Begins Work
AeroAI is developing a benchmark focused on how language models reason through aviation questions, flight-planning methods, and uncertainty.
Reducing hallucinations in aviation AI.
Why a fluent answer is not enough, and how we think about uncertainty in aviation language models.
Can a language model really understand aviation concepts?
Testing beyond terminology to see whether a model can apply ideas consistently across unfamiliar scenarios.
Ensuring accuracy in AI-assisted flight planning.
Flight planning is a chain of decisions. We examine how models handle constraints, evidence, and missing information along the way.