Why We Don't Use or Recommend AI (artificial intelligence) Transcripts

We are seeing an increasing number of AI transcriptions that we are asked to correct.  More often than not, it is more cost effective to start from scratch and create entirely human-transcribed documents.  Here's why human-transcribed data is more reliable...

AI doesn't have contextual understanding

Our team is based across the UK and we understand context, slang, idioms, and nuanced speech that AI can misinterpret or miss entirely.  Our team are specialists in deciphering accents, dialects, and poor quality audio - and we do so better than most AI systems.

AI platforms don't have the same confidentiality

Our team are vetted and bound by confidentiality agreements, providing a level of trust and security that is harder to enforce with AI.  We have a complete audit trail of your data from receipt to delivery.  Our servers are based in the UK and we have a robust retention policy and proceedure.  Some AI platforms own the data you produce, using them for analysis.  When you work with us, you retain full ownership of your data.  

AI doesn't format flexibly

Our team will format the document as they transcribe, ensuring you get a transcript that's ready for analysis - including identifying participants correctly and the anonymisation of sensitive data if required.

AI doesn't do its own quality control

AI will assume it is right and won't tell you about assumptions or limitations in its outputs.  This makes it harder for you to know where to check for inaccuracies.  Our Quality Control team independently checks each and every transcript to ensure the highest level of accuracy (currently over 98.8%), so you know that you are obtaining a true representation of the discussion.