Links

Tiktok and Ethnic Cleansing — as hardline governments rise all around the world, the whole “design policy and tools for the worst case environment” is looking a whole lot more salient. There are worse situations in the world than that of Xinjiang’s Uyghurs, but not many. Tiktok’s shadowbans and content blocks are worth learning more […]

Removing Slurs from ImageNet — The first issue is that WordNet contains offensive synsets that are inappropriate to use as image labels. Although during the construction of ImageNet in 2009 the research team removed any synset explicitly denoted as ‘offensive,’ ‘derogatory,’ ‘pejorative,’ or ‘slur’ in its gloss, this filtering was imperfect and still resulted in

Companies That Work With ICE — interesting to see technology platforms in the conversation about ethical business. I wonder who contracts to RJ Reynolds, or Shell, and whether we’ll see lists of those companies too. Matrix Methods in Data Analysis, Signal Processing, and Machine Learning — videos of lectures. (via Mat Kelcey) Sound Dataset for

The Cloud and Open Source (Tim Bray) — (language warning) There is plenty of evidence that you can be a white-hot flaming a**wipe and still ship great software. But (going out on a limb) I don’t think you can be an a**hole and be good at operations. Espresso: A Fast End-to-end Neural Speech Recognition Toolkit

Boring Technology Behind a One-Person Startup — Just dead simple things that actually work. YouTube’s Procella Database — a new query processing engine built on top of various primitives proprietary to Google. This paper stood out to me for a number of reasons: this is the SQL query engine that runs YouTube, and it has

100,000 Free Royalty-Free Faces — “generated by AI.” Beyond WEIRD Psychology: Measuring and Mapping Scales of Cultural and Psychological Distance — since psychological data is dominated by samples drawn from the United States or other WEIRD nations, this tool provides a “WEIRD scale” to assist researchers in systematically extending the existing database of psychological phenomena

Extracting Insights from the Shape of Complex Data using Topology (Nature) — three properties of topology that make it useful: coordinate-free, so you can work on data with different coordinate systems; “small” deformation invariance, so it’s resistant to some forms of noise; and it works on the compressed representations of shapes so you can use

Mirroring to Build Trust in Digital Assistants — these experiments are designed to measure whether users prefer and trust an assistant whose conversational style matches their own. To this end, we conducted a user study where subjects interacted with a digital assistant that responded in a way that either matched their conversational style, or did