The Thinking Machine

Hard fact: Automatic quality prediction based on neural models does NOT work at the segment level

LLMs Continue to Hallucinate––and accurate measurement of how much exactly, reveals a stunning picture

The work of securing AI systems will never be complete: LLMs amplify existing security risks and introduce new ones

A Thought Once Spoken Is a Lie

Why you should not base your workflow process decisions on any segment-level score (including Phrase’s new QPS)

Our Most Significant R&D Result in 2023: Edit Distance Prediction Method

Why it is important to acknowledge the lack of intelligence in “AI”

GEMBA-SQM translation quality evaluation is easy to implement as zero-shot LLM prompt … and totally useless

Translation quality evaluation is all we need

Smaller models are still better than LLMs if trained well

The AI Boogeyman: Deconstructing Harari’s Dire Warnings