# Decoding Life on Earth | Google and the Earth Biogenome Project

Source: https://www.youtube.com/watch?v=7WfnTwIM95g
Recap page: https://rapidrecap.app/video/7WfnTwIM95g
Generated: 2025-11-11T01:39:00.587+00:00

---
## Quick Overview

The Earth BioGenome Project aims to sequence the genomes of 1.8 million species across all domains of life—including animals, fish, birds, insects, plants, and fungi—to create a comprehensive genetic catalog of life on Earth, leveraging AI tools like DeepVariant, DeepConsensus, and DeepPolisher to improve accuracy and reduce costs, ultimately supporting conservation efforts like increasing the Kakapo population.

**Key Points:**
- The Earth BioGenome Project's goal is to sequence the genomes of 1.8 million species on the planet.
- Key AI tools developed by Google Research, such as DeepVariant (2018), DeepConsensus (2021), and DeepPolisher (2025), are used to improve the accuracy of genome sequencing.
- Sequencing the entire genome of a species, like the human genome, costs approximately $1,000 today, but the project aims to reduce this cost to about $100 per genome.
- AI tools help researchers assemble fragmented DNA pieces and correct errors, enabling the sequencing of entire genomes in a single day for about $1,000.
- The project directly supports conservation, exemplified by efforts to preserve the Kakapo (Strigops habroptilus), a flightless parrot whose population grew from 49 individuals in 2015 to over 200 individuals three generations later through genetic management.
- The project's scope encompasses all kingdoms of life: animals, fish, birds, insects, plants, and fungi.

![Screenshot at 0:03: Erich D. Jarvis introduces the scope of the Earth BioGenome Project, stating the goal to sequence the genomes of 1.8 million species before they become extinct.](https://ss.rapidrecap.app/screens/7WfnTwIM95g/00-00-03.png)

**Context:** This video serves as a documentary/interview piece explaining the ambitious Earth BioGenome Project, a massive scientific undertaking to sequence the entire genetic code of every known eukaryotic species on Earth. It features Erich D. Jarvis, a Professor at Rockefeller University, and Andrew Carroll, Product Lead for Genomics at Google, discussing the necessity of the project for understanding and preserving biodiversity, and the technological role Artificial Intelligence plays in achieving this goal.

## Detailed Analysis

The video outlines the monumental goal of the Earth BioGenome Project: to sequence the genomes of 1.8 million species across all domains of life, including animals, fish, birds, insects, plants, and fungi, to create a complete catalog of life. Erich D. Jarvis emphasizes the urgency, noting that many species are going extinct rapidly. Andrew Carroll of Google Research explains how AI tools are crucial for making this massive sequencing effort feasible. Google has developed several key AI milestones: DeepVariant (2018) for variant calling, DeepConsensus (2021) for assembling fragmented sequences, and DeepPolisher (2025) to reduce error rates in the final sequences. These tools help assemble the billions of small DNA pieces accurately and affordably. The cost of sequencing a human genome, currently around $1,000, is projected to drop to $100, allowing for comprehensive cataloging. A successful application is shown in the conservation of the Kakapo, a flightless parrot in New Zealand; genetic management, informed by sequencing, helped increase their population from 49 individuals in 2015 to over 200 individuals three generations later, proving the direct conservation benefit of this genomic data.

### Project Goal and Urgency

- Sequence 1.8 million species genomes
- Capture genetic code before species extinction
- Essential for understanding biology and life on Earth

### Google AI Contributions

- DeepVariant released 2018
- DeepConsensus released 2021
- DeepPolisher proposed for 2025 to reduce error rates

### Technological Impact

- AI tools improve accuracy and speed of sequencing
- Enables assembly of billions of tiny DNA pieces
- Reduces cost of sequencing to approximately $100 per genome

### Conservation Success Story

- Kakapo (Strigops habroptilus) conservation benefited from genomic data
- Population grew from 49 individuals (2015) to over 200 (three generations later)
- AI helps maximize genetic diversity in managed populations

![Screenshot at 0:01: Aerial view of a misty forest at sunrise, setting a natural, biodiverse scene.](https://ss.rapidrecap.app/screens/7WfnTwIM95g/00-00-01.png)
![Screenshot at 0:03: Erich D. Jarvis speaking in a library setting, introducing the scope of the Earth BioGenome Project.](https://ss.rapidrecap.app/screens/7WfnTwIM95g/00-00-03.png)
![Screenshot at 0:08: Visual representation of raw DNA sequencing data fragments appearing against a black background.](https://ss.rapidrecap.app/screens/7WfnTwIM95g/00-00-08.png)
![Screenshot at 0:16: Animation showing the transition from fragmented sequences to a cohesive, circular map of genetic data.](https://ss.rapidrecap.app/screens/7WfnTwIM95g/00-00-16.png)
![Screenshot at 0:22: Collage of various species \(Black-capped squirrel monkey, Northern mockingbird, Migrant hawker, English oak\) representing the diversity to be cataloged.](https://ss.rapidrecap.app/screens/7WfnTwIM95g/00-00-22.png)
![Screenshot at 0:32: Close-up of a scientist \(Andrew Carroll\) using lab equipment, indicating the physical process of sample handling.](https://ss.rapidrecap.app/screens/7WfnTwIM95g/00-00-32.png)
![Screenshot at 0:35: Microscopic view of red blood cells and another cell type, illustrating biological data being analyzed.](https://ss.rapidrecap.app/screens/7WfnTwIM95g/00-00-35.png)
![Screenshot at 0:41: Andrew Carroll explaining how AI is essential for handling the challenges of genomic data analysis.](https://ss.rapidrecap.app/screens/7WfnTwIM95g/00-00-41.png)
![Screenshot at 0:44: Erich D. Jarvis and Andrew Carroll walking outdoors, discussing the project in a collaborative setting.](https://ss.rapidrecap.app/screens/7WfnTwIM95g/00-00-44.png)
![Screenshot at 1:25: Title card displaying 'DeepVariant' \(AI Milestone I, released 2018\), marking the first major AI tool release for genome assembly/variant calling.](https://ss.rapidrecap.app/screens/7WfnTwIM95g/00-01-25.png)