check in before a long confrontation

[Ultimately_Untrue_Thought.git] / content / drafts / reducing-bias-and-improving-safety.md

diff --git a/content/drafts/reducing-bias-and-improving-safety.md b/content/drafts/reducing-bias-and-improving-safety.md
new file mode 100644 (file)
index 0000000..3cbcc7d
--- /dev/null
+++ b/content/drafts/reducing-bias-and-improving-safety.md
@@ -0,0 +1,19 @@
+Title: "Reducing" Bias and Improving "Safety"
+Date: 2022-07-19
+Category: commentary
+Tags: artificial intelligence
+Status: draft
+
+https://openai.com/blog/reducing-bias-and-improving-safety-in-dall-e-2/
+
+> Today, we are implementing a new technique so that DALL·E generates images of people that more accurately reflect the diversity of the world’s population. This technique is applied at the system level when DALL·E is given a prompt describing a person that does not specify race or gender, like "firefighter."
+
+https://www.reddit.com/r/dalle2/comments/w1mflp/gender_bias_gone/igmevpj/
+https://twitter.com/waxpancake/status/1549076996935675904
+https://twitter.com/rzhang88/status/1549472829304741888
+https://twitter.com/cis_female/status/1549526889420947456
+
+> it doesn't seem to want to give me a male nurse
+https://twitter.com/robertskmiles/status/1549559401622691840
+
+https://www.lesswrong.com/posts/pRkFkzwKZ2zfa3R6H/without-specific-countermeasures-the-easiest-path-to