So there really has to be this human-in-the-loop concept to actually look at it and say 'Is this behaving the way we thought it should be behaving?' And if it's not, let's figure out when it went off the rails, so to speak. - Maribel Lopez
Second, we really need to think about representative data. For the most part, models are not trained on the world we'd like to see, let's just say. There's missing data, underrepresented data, and data that may have changed. Think of historical records or legal precedents, all those things change - Maribel Lopez
------------
Maribel packs a punch of considerations we need to keep in mind when building, managing, and using algorithmic models. Not only are they imperfect to begin with, they're changing all the time.
A small sample of a far-ranging conversation with one of the brightest minds in our industry.
Maribel Lopez: Contextual Intelligence, Ethics and Well-Being | Work 20XX with Jeff Frick
YouTube -
- transcript and show notes - https://www.work20xx.com/episode/maribel-lopez-contextual-intelligence-ethics-and-well-being-work-20xx-08 - WebexAhead - https://webexahead.webex.com/contextual-intelligence-ethics-and-well-being/ - This episode of Work20XX brought to you by Webex by Cisco
#ai #models #algorithms #explainableai #blackbox #data #work #underrepresenteddata #changingdata #ethings #syntheticdata #fow #lopezresearch #webexahead #work20xx #webex #cisco #podcast #interview #captioned #cxo #change #analytics #dynamic #research #intelligence
Share this post