Advertisement

The ‘Creepy’ Potential of AI Notetaking

Article main image
Oct 18, 2018

I first seriously looked at automated transcription 10 or 15 years ago, when Dragon Naturally Speaking was the leading product. For my purposes, the tech wasn’t quite ready then, but now, thanks to advances in AI, transcription is a genuinely useful tool. It’s even built into Google Docs, Windows, and  iOS as a standard feature.

The question for HR is whether AI transcription should become a standard tool for interviews or meetings. One new tool is Voicera’s Eva which uses several different AI engines to maximize the quality of transcription. The tool links to your calendar so you can just turn on transcription any time it’s called for — it’s easy enough that you can use it all the time.

In my limited experience the transcript makes sense if you were at the meeting or interview, but it wouldn’t stand on its own as a record without editing or going back to the audio recording. Given how easy it is to have a discussion transcribed, it’s tempting to have this turned on for any important meeting just so you could go back, if needed, to clarify something.

More intriguing is that Eva pulls out a summary of the meeting by looking for key phrases like “Action Item” and “We decided.” This works better if you explicitly use the key phrases whenever you want something to appear in the summary, but that’s not hard to do.

This last feature — and what it says about the future — is why HR should begin paying attention to this technology. If an AI can regularly listen to interviews and meetings, and start making sense of them, then it might provide insights from across many conversations. Imagine an AI assessing conversations across the organization for signs of increasing stress or even noticing that different teams were discussing similar issues.

There really is no obvious ceiling on what could come from having someone (if we dare call an AI “someone”) keeping an eye on all the conversations happening in the organization and tracking any indicators management feel should be monitored. Yes, this could get creepy, but we are already deep in a world of tech creepiness, and if there is a genuinely useful application then it will just be a matter of adding some safeguards, not eschewing the technology all together. It’s best to begin thinking about this now.

What is interesting

  • If we tried out a technology a few years ago we may think we understand it, yet it’s easy to be wrong because the technology is always changing.

What is really important

  • We’ve reached a point where an AI engine can easily transcribe any conversation we want transcribed. It’s not perfect, but it’s good enough to be useful.
  • The summarization of key points by the AI is both useful and a signal of where the tech is going.
  • There is no end to what this technology might do in the next 5 or 10 years. In fact, novel applications of the technology — even as it exists today — are waiting to be discovered.
  • More than ever HR needs to keep an eye on, and play with, new technologies so that it stays up-to-date on what’s possible.