Oct 26, 2017

Axios Science

By Alison Snyder
Alison Snyder

Thanks for your feedback about Axios Science and what you are reading in the daily stream. Please keep it coming. You can send me a note at alison@axios.com or reply to this email.

Computers are learning to recognize letters like we do

One of the ways computers distinguish humans from robots is with CAPTCHAs — that little box with a weird letter combination at the bottom of your online ticket or other transaction. Researchers report they've now trained a computer to solve CAPTCHAs using less data than other AIs by borrowing the human brain's approach to the problem.

The big picture: This isn't about cracking CAPTCHAs, but a much larger effort to create AIs that use principles of the human brain to solve visual tasks — like recognizing a cat from a dog with not a lot of examples to go on. AI pioneer Geoffrey Hinton recently told Axios he suspects the current approach of giving computers lots of rules and labeled data is limited, and that researchers should look again to the brain to make material advances in AI. This is a step in that direction.

Read my full story here.

Axios stories to spark your brain
  • Robotic surgery may not be worth it, writes Eileen O'Reilly. The outcomes for at least two types of surgeries aren't significantly better than conventional surgeries to warrant the high cost of the operations, according to two new studies.
  • Traffic solution: Steve LeVine on Pittsburgh's new traffic system that uses cameras, radios and radar to change lights at an intersection in order to optimize the flow of traffic.
  • Gold: Paul Sutter writes that new evidence about a source of heavy elements in the universe also raises questions about how much is made in stellar explosions versus collisions.
The next generation of gene-editing tools

The gene-editing tool CRISPR promises major scientific advances, but it has been held back by concerns about the precision of its editing and the ethics of making permanent changes to DNA. Now, two new papers describe tools that potentially get around these technical and ethical concerns, writes Axios' Erin Ross.

Why it matters: CRISPR gene editing has the potential to cure many genetic illnesses. Its potential is only just being tapped, but so far it's been used to engineer better tomatoes, change the colors of butterfly wings, test for diseases and has even been used to edit human embryos. The main roadblocks have been technical and ethical problems — which could be partly addressed with new techniques that allow for more accurate and less risky gene editing.

Read the rest of Erin's post here.

What we're reading elsewhere
Something wondrous

In 1996, an expedition of ornithologists John O'Neill, Dan Lane and Andy Kratter headed into central Peru. They were en route to the cloud forest of the eastern Andes when they stopped on a ridge top and spotted a bird called a manakin. Its yellow breast feathers threw the researchers. The manakin subspecies in the area didn't have them but there were, O'Neill knew, yellow-breasted manakins over a thousand miles away in Venezuela.

More than twenty years passed and the researchers finally got their hands on a recording of the Venezuelan bird's song. Differences in the songs correlate with genetic differences, so the researchers used the recording instead of DNA analysis as a measure of similarity between the birds. The song of the bird from Venezuela was different from that of the bird they'd discovered years ago in Peru. For one, the yellow-breasted bird in Peru doesn't have undertones.

That, along with some differences in the sounds their wings make and their plumes, led the researchers to recently ask the committee that oversees the taxonomy of South American birds to deem the manakin a separate species. They propose naming it Machaeropterus eckelberryi after Don Eckelberry, a bird illustrator and mentor to Lane and O'Neill. Kratter, now the collection manager of birds at the Florida Museum of Natural History, says the bird would be known in English as the Painted Manakin.

Alison Snyder