[Career review] “Data collection for AI alignment” by Benjamin Hilton

Many proposals to reduce risks from misaligned AI may need a lot of high-quality data from people that can be used to train machine learning models. Narrated by AI. ---Outline:(00:47) Sometimes recommended -- personal fit dependent(00:59) Review status(01:04) Why might becoming an expert in data collection for AI alignment be high impact?(02:58) What does this path involve?(04:29) Avoiding harm(05:21) Example people(05:25) Long Ouyang(06:31) How to predict your fit in advance(07:17) How to enter(09:09) Find a job in this path(09:21) Want one-on-one advice on pursuing this path?(09:42) Learn more about data collection for AI alignment(10:09) Learn about other high-impact careers --- First published: May 11th, 2022 Source: https://80000hours.org/career-reviews/alignment-data-expert --- Narrated by TYPE III AUDIO.

Episode metadata supplied by the publisher feed · Published May 11, 2022

Embed this episode

Attribution link and audio player

NOW PLAYING

[Career review] “Data collection for AI alignment” by Benjamin Hilton

0:00 10:37

1×

No transcript for this episode yet

We transcribe on demand. Request one and we'll notify you when it's ready — usually under 10 minutes.

Share this episode

Similar Episodes

No similar episodes found.

Similar Podcasts

TED Talks Society and Culture Thought-provoking videos about life and being human, with ideas from business leaders, psychologists and researchers speaking onstage at the TED conference, TEDx events and partner events around the world. You can also download these and many other videos free on TED.com, with an interactive English transcript and subtitles in up to 80 languages. TED is a nonprofit devoted to Ideas Worth Spreading.

Frequently Asked Questions

How long is this episode of 80,000 Hours - Narrations?

This episode is 10 minutes long.

When was this 80,000 Hours - Narrations episode published?

This episode was published on May 11, 2022.

Can I download this 80,000 Hours - Narrations episode?

Yes. Use the download control on the episode player to save the publisher-provided media file.

URL copied to clipboard!