What Adam is Reading - Week of 2-20-23

Week of February 20, 2023

 

This week, we are using our kids' mid-winter school break to visit family in Reno, Nevada.   There are no direct flights from Maryland, so we spent many hours on planes.  Despite the long hours in public together, I am happy to report the real-time feedback from my teenagers reflected only "minimal, but some embarrassment due to making friends with strangers and overly engaging flight attendants in conversation."  Feedback is a gift, and with teenagers, every day is Christmas.

 

---

The U.S. is experiencing stable average case rates and 380 deaths per day (on average) over the last 14 days). 

 

N.Y. Times Tracker

https://www.nytimes.com/interactive/2021/us/covid-cases.html

 

------

Zania Stamataki, an immunologist/virologist in the U.K., offered a thorough analysis of some recent articles looking at the immune protection conferred by COVID infections - i.e., how do post-COVID vs. vaccination-induced immunity compare?

"While a natural infection may offer equivalent protection to vaccination, this is not to say you should seek to become infected.  SARS-CoV-2 remains a dangerous and unpredictable virus that can [sometimes] cause damaging effects that linger long after recovery.

The authors suggest a person's previous infection status and timing should be considered alongside their booster vaccinations to predict [maximal, sustained] protection."

https://theconversation.com/how-much-immunity-do-we-get-from-a-covid-infection-large-study-offers-new-clues-200044

 

I found this article thanks to Eric Topol.  It is an excellent review of the interplay between COVID and individuals with chronic comorbidities.  Specifically, the authors emphasize the complex interplay of coronavirus infection and underlying diseases.    This discussion gets to the whole (IMO, absurd) question of "did someone die FROM COVID or WITH COVID?" The bottom line is don't get COVID, doubly so if you have underlying health issues.

https://www.nature.com/articles/s41591-022-02156-9

commentary

https://twitter.com/EricTopol/status/1626993050764574722

 

 

Medical Trends and Technology

 

It's time for an update on the burgeoning technology of brain-computer interfaces (BCI).  Synchron, partially funded by Bill Gates and Jeff Bezos, is starting broader clinical trials of its neurovascularly-implanted device, allowing humans to manipulate electronic devices with their minds.  While the handful of test subjects have been disabled individuals, we are marching down the augmented human path. 

https://www.cnbc.com/2023/02/18/synchron-backed-by-bezos-and-gates-tests-brain-computer-interface.html

Fortunately, we have years of science fiction to guide us through the various trade-offs this technology presents.  I refer you to Netflix's Love, Death, and Robots as a good series of stories exploring this topic.

 

Neuralink is Elon Musk's company working on this as well but has a different strategy - a chip below the skin, with transcranial wiring implanted by an 8-foot tall robot.  Seriously.

https://www.businessinsider.com/neuralink-elon-musk-microchips-brains-ai-2021-2#the-second-is-a-robot-that-could-automatically-implant-the-chip-3

 

 

Infographics

Last Friday was National Battery Day.  Learn about lithium-ion chemistry:

https://www.compoundchem.com/2019/10/09/2019nobelchemistry/

And learn about lithium battery recycling:

https://www.science.org/content/article/millions-electric-cars-are-coming-what-happens-all-dead-batteries

 

 

Things I learned this week

 

I did not know about the photography trend of "Accidental Renaissance" Photos.  I may have to take my family to Italy all over again.

https://petapixel.com/2021/05/27/how-to-create-accidental-renaissance-photos-light-and-composition/

 

I have several perpetual, low-grade anxieties about my bird feeder.  Am I driving bird obesity?  Am I fostering an unnatural species co-mingling (and the spread of disease)?  Most of all, am I creating undue dependence on humans?  New data suggest bird feeders are more Frito-lay than Ruth's Chris.  In only the second article looking at the impact of bird feeders on feeding behaviors, Chickadees (monitored under controlled conditions) do not rely on bird feeders as a primary food source but as a snack.  

https://www.popsci.com/science/wild-birds-dont-rely-on-feeders/

The cited article is from the Journal of Avian Biology.

https://onlinelibrary.wiley.com/doi/epdf/10.1111/jav.02782

 

Israeli biologists found that some flowers, when exposed to real or pre-recorded bee buzzing, make sweeter nectar - within 3 minutes of "hearing" the sound.

https://www.newscientist.com/article/2189875-flowers-hear-bees-and-make-sweeter-nectar-when-theyre-buzzing-nearby/

and

https://www.biorxiv.org/content/10.1101/507319v1

 

Similarly, a study demonstrates that gardeners reading to their tomato plants saw increased growth.  Female voices induced more growth than male voices.

https://sites.psu.edu/siowfa15/2015/10/21/can-talking-to-plant-help-it-grow/

 

 

Living with A.I.

 

I am starting to see integrations of ChatGPT into a variety of tools. 

 

A loyal reader shared Doximity's physician-focused ChatGPT interface that offers pre-typed questions for doctors - everything from referral letters to insurance denial appeals letters to patient instructions.

https://www.doximity.com/docs-gpt

 

More broadly, various developers are figuring out how to integrate A.I. into native operating systems.  Here is a developer using the MacOS language APIs to permit the use of ChatGPT in any app on the Mac.

https://twitter.com/ronithhh/status/1627118300194013186

 

Of course, there are a variety of fun discoveries and imperfections from Microsoft's A.I. implementation in Bing: https://www.digitaltrends.com/computing/chatgpt-bing-hands-on/

 

A.I. is only as powerful as the data set it is trained on and can access.  Carl Bergstrom (biology professor and author of Calling Bullshit) highlighted an emblematic case study on his website.  In 2016 researchers used a skewed training set to develop an algorithm for predicting who would commit a crime by facial features alone.  Bergstrom highlighted how flawed their logic was and, thus, how the output was biased.

https://www.callingbullshit.org/case_studies/case_study_criminal_machine_learning.html

 

Watch for the increasing number of articles highlighting similar flawed output from emerging A.I. tools.  For instance, the internet is a biased data source.  At some point, A.I. tools are simply regurgitating the most statistically probable bits of our knowledge, albeit rapidly and organized.

 

A.I. art of the week

"A robot reading to a plant, computer art"

https://labs.openai.com/s/zKnU6Qhq8uYdH9DSJ9gZHmUf

 

 

Clean hands and sharp minds,

Adam

Comments