You’re not sending an e-mail. You’re training a model
Privacy, employee data and the demands of AI.
Sign up now: Get ST's newsletters delivered to your inbox
Legal environments notwithstanding, turning employee interactions into AI feedstock imposes other costs.
PHOTO: UNSPLASH
The Economist
One way to think about AI note-takers is as a productivity tool for humans. You might be unable to attend a gathering but you can still find out what happened by reading a summary later. You can receive a set of specific action points rather than operating on the vague assumption that Doris will follow up on something or other.
But another way to think about AI note-takers is as a data-gathering exercise for other machines. Transcripts can be scoured for previously unseen patterns: how much individuals are contributing; where coaching is needed. And as offhand remarks are turned into permanent records, more employee communication becomes available as training material for AI models.