[Dcouncil] AAAS Recap - DEIA Lens

[Rebecca Anderson]

Hi team-
Thanks for letting me share some of the research I heard during the AAAS Conference. Attached is the slide deck I made for the DEIA Council. Because I usually don’t do a whole lot of words on slides, here is some context but please holler at me if you want more details.

The overall takeaway from the conference was that collaboration, connection and diversity makes us stronger. Breaking down silos or walls allows us to be better humans and researchers.

Here are some of the sources for your reference. There may be a paywall for the NYT articles and if you can’t access the piece, please let me know and I will see if there is a workaround.

Slides 3-4 – How do we know AI can sometimes be racist?
There was a study conducted where images of two fraternities shown side by side yielded different targeted ads. Fraternities from HBCUs saw higher incidences of targeted ads that advertised high credit card interest rates and ads with the phrase to “expunge an arrest record.” One of the reasons cited for this was due to “traumatized data”:

  *   Traumatized data – data that has misconceived notion, carries a memory that impacts value, norms and association. We don’t do a good job fixing this because the people impacted are not sitting at the table.
  *   https://www.brookings.edu/articles/the-u-s-can-improve-its-ai-governance-strategy-by-addressing-online-biases/

Slides 5-6 – The problem with facial recognition technology:
If AI is relied on too heavily, without proper human intervention, the wrong people can be convicted. Randal Quran Reid (man on the right) was arrested in GA for a crime in a state that he had never even been to. Porcha Woodruff, (the woman on the left) was arrested when she was 8 months pregnant in front of her children for a crime she did not commit. She went into premature labor due to dehydration sitting in a holding cell.

  *   https://www.nytimes.com/2023/08/06/business/facial-recognition-false-arrest.html
  *   https://innocenceproject.org/when-artificial-intelligence-gets-it-wrong/

Slides 7-8 – The problem with predictive text:
Traumatized data can influence search results and seemingly innocent search words can be associated with certain phrases due to a content moderation system.

  *   https://www.wbur.org/hereandnow/2021/09/30/safiya-noble-internet-research
  *   https://nyupress.org/9781479837243/algorithms-of-oppression/

Slides 10-11: Tackling misinformation:
The retracted article has been cited 67 times. One of the most dangerous things that can happen is when well-known outlets cite these articles. This was published in February and this was the article cited was this retracted one. If you just scroll down to the sources and follow-through by clicking the link, it shows up as retracted.

  *   https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10102036/
  *   https://www.nytimes.com/2024/02/02/opinion/transgender-children-gender-dysphoria.html

Slide 12: Possibilities for conferences
Slide 13: I liked the signage for accessibility purposes.
Slide 14: I added a snippet from Johnathon Huff’s recent COO Operations meetings as I thought it was timely and supported the overall message of the conference.
Slide 15: A photo I snuck of Cassandra doing her thing during the conference.

Thanks, R

Rebecca Anderson
Internal Communications Specialist
DEIA Council Co-Chair
Thomas Jefferson National Accelerator Facility
Jefferson Science Associates, LLC
O: 757.269.6809 | C: 757.679.7575
duckett at jlab.org<mailto:duckett at jlab.org>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.jlab.org/pipermail/dcouncil/attachments/20240322/4b8d3c04/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 3.20_randerson_movingtowardsciencewithoutwalls.pdf
Type: application/pdf
Size: 17278729 bytes
Desc: 3.20_randerson_movingtowardsciencewithoutwalls.pdf
URL: <https://mailman.jlab.org/pipermail/dcouncil/attachments/20240322/4b8d3c04/attachment-0001.pdf>