Smash Notes
Why Silicon Valley Is Hiring Bird Experts
A few years ago, reporter Sarah McBride noticed that a top engineer at Twitter was also an expert on the brains of birds. Then, more and more, she started seeing that many top tech companies have bird brain experts in their highest ranks—that includes Apple, Google, Intel and a secretive startup founded by Elon Musk. This week on Decrypted, Sarah and…
Notes
-
"The zebra finch is the most common estrildid finch of Central Australia and ranges over most of the continent, avoiding only the cool moist south and some areas of the tropical far north." - According to Zebra Finch description on Wikipedia.
-
At the core of his exploration, he wanted to explore why a task seemed so easy for a child, but was difficult for a robot.
-
He has honed in on why zebra finches learn to sing. It is a means to an end as he seeks ways to ultimately gain a better understanding of how the human brain functions.
-
The old title of this note says birds learn to "sign." On Decrypted, Sarah McBride and Brad Stone are talking about singing. Tim Otchy, a neuroscientist at Boston University, studies zebra-finch brains for general rules of how brains work. Only the males sing. Baby birds imitate a father's song, badly at first and better a month later. The songbird is one of the systems where the brain regions are best mapped, so researchers can ask precise questions about brain activity and behavior, including how a human learns. Otchy's lab uses a benign virus so neurons glow, then implants tiny microscopes. McBride says tech companies care because the labs produce huge data sets, and because the questions sit inside artificial intelligence.
-
Scientists working on animals studies have a lot of interesting data, and tech companies love access to data. Additionally, the research provides useful insights into artificial intelligence.
-
Human brains are too complicated to study, nevermind the ethical implications of doing it, while animal brains a lot simpler and we have set principles and technologies to study them.
-
On Decrypted, Sarah McBride tells Brad Stone that neuroscience is starting to shape how companies design AI, so it is a good moment to be a neuroscientist. People who had spent a career in universities were getting offers in Silicon Valley. Google, Apple, and Amazon were hiring them, including the zebra-finch researchers this episode follows. Intel will not say how many people it has in AI. McBride says it is a lot, and that many of them come from this animal-neuroscience background. The pitch she describes is machines that behave in smarter, more human ways. A self-driving car that has to decide quickly was already one of the jobs.
-
Amir Khosrowshahi is a computational neuroscientist by training, and he says that a lot of AI technologies can benefit from learning how humans perform tasks that computers don't yet know how to do well.
-
Decrypted puts the question to Intel. Tyler Lee, a UC Berkeley PhD hired by Amir, another Berkeley-trained computational neuroscientist, works on virtual assistants. Much of it is context. In a car, knowing the speaker is the driver or the passenger makes the request easier, because a driver is more likely to ask for directions. Lee says that resembles his zebra-finch work. A bird recognizes the kind of call first, then who is calling: a mate, family, or a threat. Amir says his team has also studied flies, rats, locusts, and worms. Those simple animals still do things neural networks cannot simulate.
-
On Bloomberg's Decrypted, Sarah McBride and Brad Stone ask when zebra-finch research shows up in products. The first answers are on the phone: voice unlock, and noise reduction on calls and video calls. Past gadgets, the same animal work feeds medicine, including how to stop Parkinson's tremors and how to handle prosthetic limbs. The further case is Neuralink. Tim Gardner left the Boston University lab, later run by Tim Otchy, to work for Elon Musk. Ashlee Vance describes a brain-machine interface as a two-way channel. Implants that help people hear, or that calm tremors, already exist. The version being chased went further, including a mesh on the brain.
-
Neuralink could enable people to have a superhuman cognition, where's one could download an entire library into their brain, or learn a language in an instant.
-
Brad Stone asks Sarah McBride, at the end of Bloomberg's Decrypted, whether it is too early to hear a zebra finch inside Alexa, Siri, or Google's voice assistant. McBride says some of the research is already visible, in voice recognition and in better audio quality. She thinks the biggest consequences are still ahead. The episode's reporting had followed zebra-finch neuroscientists out of university labs and into companies such as Intel, Google, Apple, and Amazon, including Tim Gardner's move from Boston University to Neuralink. McBride's answer is the cautious one. You can see the early applications. The larger transfer from bird brains to machines had not arrived yet.
Related notes
- Is There Any Potential Research That Focuses on Creating an AI Neural Network That Can Be Applied to the Brain
- How Will AI Affect the Ability for Humans to Learn, Build, and Create
- How Do You Think About Neuralink in the Future?
- Are the Scientists Trying to Make Tools to Solve What Humans Can’t Solve or Replicate Human Intelligence
- Is There a Connection Between Neural Networks and the Human Brain
- Why Does a Researcher Join an Artificial Intelligence Company?
Related episodes
- Silicon Valley Is Making Gasoline Out of Thin Air
- Kirill Zubovsky: the World of Podcasting
- Dr Raia Hadsell - Artificial Intelligence researcher
- How Humans are Artificially Manipulating Weather | KMP Ep.57
- ‘This Is Nuts.’ An OpenAI Insider Explains Why He Quit.
- #521: Dr. Andrew Huberman — a Neurobiologist on Optimizing Sleep, Performance, and Testosterone