As we being our countdown to Halloween, let’s take a look at flawed training datasets, disturbing developments in AI, and how we can fight back against unethical AI.
Sign up to get the full-length version of this content delivered to your inbox every Monday!
Here are some datasets that were recently taken down but refuse to go away.
MIT apologizes, permanently pulls offline huge dataset that taught AI systems to use racist, misogynistic slurs
80 Million Tiny Images has thousands of images labelled with racist slurs for Black and Asian people, and derogatory terms used to describe women – needlessly linking everyday imagery to slurs and offensive language, and baking prejudice and bias into future AI models.
An online image database will remove 600,000 pictures after an art project revealed the system’s racist bias
ImageNet (created by Stanford researchers), a popular online database of images, removed 600,000 pictures of people from its system after an art project revealed the depths of the racial biases of the system’s artificial intelligence.
Facial recognition datasets are being widely used despite being taken down due to ethical concerns. Here’s how.
h/t Arvind Narayanan @random_walker
DukeMTMC dataset of videos was used in 135 papers published after it was taken down in June 2019. MS-Celeb-1M dataset was removed by Microsoft in 2019 after receiving criticism. The dataset lives on through several derived datasets, including MS1M-IBUG, MS1M-ArcFace, and MS1M-RetinaFace.
OkCupid Study Reveals the Perils of Big-Data Science (published in 2016)
h/t Casey Fiesler, PhD, JD, geekD @cfiesler
For those concerned about privacy, research ethics, and the growing practice of publicly releasing large data sets, this logic of “but the data is already public” is an all-too-familiar refrain used to gloss over thorny ethical concerns.
Horror-ific News
Facebook Seeks Shutdown of NYU Research Project Into Political Ad Targeting
h/t emily bell @emilybell
Facebook Inc. is demanding that a New York University research project cease collecting data about its political-ad-targeting practices, setting up a fight with academics seeking to study the platform without the company’s permission.
Palantir to Help US Track Covid-19 Vaccines – WSJ
h/t LouisColumbus @LouisColumbus
Data-mining company Palantir Technologies Inc. is helping the federal government set up a system that will track the manufacture, distribution and administration of Covid-19 vaccines, state and local health officials briefed on the effort said.
Google AI Tech Will Be Used for Virtual Border Wall, CBP Contract Shows
h/t Meredith Whittaker @mer__edith
After years of backlash over controversial government work, Google technology will be used to aid the Trump administration’s efforts to fortify the U.S.-Mexico border, according to documents related to a federal contract.
Politicians have made an algorithm to fix the housing crisis. It’s bad.
h/t Your Algorithm Doesn’t Know Me @algorithm_me
Another “mutant algorithm” haunts the halls of power: a proposed new formula to decide where to build 300,000 new homes and solve England’s housing crisis.
Sign up to get the full-length version of this content delivered to your inbox every Monday!