š¦“More promising AI uses in radiology and it is a "moral imperative" to consider killing people with AI
š± Venus flytraps get some robo-attachments in the latest tech advancement
The Good
Two recent stories, both in Wired, showed some neat potential uses of AI-powered radiology.
One story covers a new algo that is better at predicting breast cancer risk than previous models. And since women have been skipping breast cancer checks during COVID-19, this algo can help prioritize the most in need to come in and get checked. Super beneficial during the pandemic, but a helpful tool even when things calm down.
The other study found that after being trained on arthritic patientsā reports, an algo did better job at recognizing the level of pain experience by black patients. Doctors have historically and currently believed that Black patients can withstand pain more than white Americans, leading to uneven and unequal medical treatment. While we have to worry about algos not being trained on a diverse dataset and causing different problems to creep in, this is a demonstration of how tech can be used to mitigate human biases.
The Bad
Iām not sure if this headline can be improved: Congressional Panel: We Have a 'Moral Imperative' to At Least Consider Building a Terminator.
The congressional advisory panel, National Security Commission on Artificial Intelligence, has former Google CEO Eric Schmidt at the helm. The group was tasked on whether or not to use AI-powered weapons in and came down to on it was aĀ āmoral imperativeā to consider it.
Listen. A discussion about autonomous weapons is important. Iām not a moral philosophy professor (and what do you know, doesnsāt seem there were any on the panel either), but calling it a āmoral imperativeā is not striking the right tone. There is a similar argument about why NOT discussing it is the āmoral imperative.ā
Many of the other suggestions that the group came up with were fine, and its a shame that they werenāt able to make a strong statement about the perils of autonomous weapons instead of some Cold War-inspired postering.
More News
Maybe we should have been less worried about a man/machine cyborg, and more worried about a plant/machine one.
Google is shutting down the companyās autonomous floating cell tower balloon project.
Amnesty International is creating a crowdsourced map showing every face rec tech-enabled camera in NYC.
A Q&A with a cosmologist researching a concept for a āself-driving telescopeā and more ethical algo development.
Sure Facebook, Twitter, and YouTube algos are bad at suppressing misinformation, but turns out, so are Amazonās!
SpotifyĀ got a patent for tech that will try to figure out what music to recommend by the usersā speech and background noise.
AI could help smokers control their cravings and even quit smoking (or in this first instance, vaping).
A thinky-thought piece about what our focus on āpredictionā technology means for the future and how our understanding of the brain brought us here.
Men might to get to thank the AI gods that checking for prostrate cancer might not include a probe or blood sample in the future.
***
Until moral philosophy professors get their due,
Jackie