Ben Packer

Data Science and AI at Recidiviz. Previously: Responsible AI at Google, Principal Data Scientist at Opower.

[Ben Packer @packer_ben]

My post with Yoni Halpern, Mario Guajardo-Céspedes, and @mmitchell_ai on #MLFairness and bias in text models: https://t.co/m9v1QCgX9m

[Google for Developers @googledevs]

To stop problematic biases from appearing in our ML models, we need to be able to identify them. In this post from the Google AI team, we detail example case studies where gender bias exists in the input text embeddings.

Find out more on the blog ↓

https://t.co/RzUo2DvaSd

[Ben Packer @packer_ben]

@KLdivergence I remember my advisor teaching this with an eyeroll and a "sorry, this is from a by-gone era, this is just what they call it" but I'm not sure if they teach it with different terminology these days. @alexdamour?

[Ben Packer @packer_ben]

@KLdivergence In the graphical models literature, graphs are considered "moral" if the parent nodes are connected ("married"): https://t.co/RRpUmOiupj

[Ben Packer @packer_ben]

@ravenswoodfhn Is there a Spanish version of this flyer to share?

[Krishnaram Kenthapadi @kkenthapadi]

Attending #AAAI2021? Please join our tutorial on #ResponsibleAI in Industry [8:30am PT tomorrow]: https://t.co/oxhLa3RBtr [w/ @packer_ben Mehrnoosh Sameki @phenomenashlie].
Slides: https://t.co/iR45wFyLj3
#ExplainableAI #Fairness #Privacy #MachineLearning #CaseStudies @RealAAAI

[Ben Packer @packer_ben]

@rajiinio @KLdivergence @ruchowdh @Aaron_Horowitz @AlexCEngler Exactly. "The spreadsheet messed up" doesn't sound like as good of an excuse.

[Ben Packer @packer_ben]

@mark_riedl Sota on a sofa?

[Sergio (🌈)🕊️#PalestineLivesMatter @sguada]

I support @timnitGebru, she is an amazing researcher and a force for real change.
She has changed @googleAI and myself so much and showed everyone how real leaders lead.

[Ben Packer @packer_ben]

@Taniel Is it still worth donating to Vance Keyes? You can donate through his website here: https://t.co/dA7I0Oj4bY

[Ben Packer @packer_ben]

@sashaperigo @TheLeagueSF @DemSocialists @DSAEastBay @DSA_SF @DSA_LosAngeles Do you know of any Spanish-language voter guides for San Jose/Santa Clara County? Looking for a friend, and I haven't been able to find a trusted endorsement list.

[Ben Packer @packer_ben]

@AngieBEvans "By Appointment Only" in case you thought you could just wander in to the shack as a walk-in and expect to be given the tour

[Ben Packer @packer_ben]

@kath_rothschild I read Are You There God? It's Me, Margaret as an 8 year old boy and... I did not understand what they were talking about

[Ben Packer @packer_ben]

@sashaperigo The peninsula chapter of YIMBY Action @penforeveryone endorsed Prop 21 (and Prop 16) fwiw

[Ben Packer @packer_ben]

@KLdivergence Jk I know the feeling, carry on

[Ben Packer @packer_ben]

@KLdivergence Why are you using weapons against your toddler?

[Ben Packer @packer_ben]

@paulg This is why so much of Responsible AI work is focused on actual actions and real-world effects of AI, not just on expressions and internal representations.

E.g. @PopTechWorks's Automating Inequality, @KLdivergence's criminal justice work, and many more. https://t.co/okvlhyeM7C

[Ben Packer @packer_ben]

@paulg This is why so much of Responsible AI work is focused on actual actions and real-world effects of AI, not just on expressions and internal representations.

E.g. @PopTechWorks's Automating Inequality, @KLdivergence's criminal justice work, and many more. https://t.co/okvlhyeM7C

[Paul Graham @paulg]

People get mad when AIs do or say politically incorrect things. What if it's hard to prevent them from drawing such conclusions, and the easiest way to fix this is to teach them to hide what they think? That seems a scary skill to start teaching AIs.

[Ben Packer @packer_ben]

@zehavoc @pasini_t @yoavgo @gdb @GDP Yes, most/all do have issues like these.

[Ben Packer @packer_ben]

@yoavgo @gdb Any exploration of stuff like this? https://t.co/77nWOImD1s. Not trying to pile on, just might be useful in light of your comment about this being a "pattern matcher," and what GPT-3 is and isn't.

[Jerome Pesenti @an_open_mind]

#gpt3 is surprising and creative but it’s also unsafe due to harmful biases. Prompted to write tweets from one word - Jews, black, women, holocaust - it came up with these (https://t.co/G5POcerE1h). We need more progress on #ResponsibleAI before putting NLG models in production.

[Ben Packer @packer_ben]

@pasini_t @zehavoc @yoavgo @gdb @GDP FYI https://t.co/77nWOImD1s

[Jerome Pesenti @an_open_mind]

#gpt3 is surprising and creative but it’s also unsafe due to harmful biases. Prompted to write tweets from one word - Jews, black, women, holocaust - it came up with these (https://t.co/G5POcerE1h). We need more progress on #ResponsibleAI before putting NLG models in production.

[Ben Packer @packer_ben]

@BridgetMarie I had no idea that John Lewis wrote this autobiographical graphic novel about the Civil Rights Movement: https://t.co/HRtrLVlqPv (it's also available at our local library).

Getting this for my daughter, we'll see if she's old enough. https://t.co/oH4KSekEyj