🔍 Read the full analysis: From AI To Kelp Recovery: What Google And CSIRO Have Shown So Far on ThorstenMeyerAI.com
Get the latest gadgets delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
Google says it and CSIRO used kelp-specific AI tools and new Australian reference genomes to analyze 600 wild giant kelp samples from 40 Tasmanian coastal sites. The goal is to identify survivors for selective breeding, but the sample analysis has not yet been published in a paper or dataset, and the bred kelp has not been tested through warm summers at sea.
Google said November 12 that it and Australia’s national science agency, CSIRO, used AI adapted for kelp DNA to analyze 600 wild giant kelp samples from 40 sites along Tasmania’s coast. The work is intended to help scientists select surviving plants for selective breeding, but the sample analysis has so far been described in Google’s announcement rather than a published paper, and whether offspring can survive warmer seas remains untested.
According to Google’s post and accompanying video, researchers used a version of DeepVariant, a tool whose standard models are trained on human data, adapted to analyze kelp genomes. The genetic analysis is meant to help identify differences among wild plants and guide which survivors scientists cross in the lab. Google reported that the 600 specimens came from 40 Tasmanian coastal sites and said the remaining kelp patches are genetically isolated. That finding has not yet been backed by a paper or dataset identified in the source material.
The work builds on two Australian giant kelp reference genomes produced with Google’s open-source tools. A study published in BMC Genomics in June 2026 reported genomes from plants collected in Victoria and Tasmania. The researchers placed 98–99% of each genome into 35 pseudo-chromosomes. The two Australian genomes differed from each other by 0.2%, while each differed from a Californian reference genome by 1.5%, according to the study’s abstract.
Google’s video says DeepConsensus and DeepPolisher helped correct errors in raw genetic reads used to build the reference genomes. Those genomes give scientists a better local basis for comparing Australian samples than relying solely on a Californian plant. The project does not use AI to plant kelp or directly measure heat tolerance: laboratory tests provide biological information that researchers can then relate to DNA data.
Giant kelp · Tasmania · CSIRO with Google AI
From the reef to the lab and back: where the AI fits
Five steps, as described by CSIRO and Google. The AI works in steps 2 and 3; people do the diving, the lab work and the planting.
1Reef
Collect wild kelp
600 wild specimens from 40 sites along the Tasmanian coast, according to Google.
2Lab + AI
Build a local reference
Long-read DNA plus Google’s DeepConsensus and DeepPolisher: two Australian reference genomes, published in BMC Genomics.
AI step3AI
Read each plant
A version of DeepVariant trained for kelp finds where each specimen’s DNA differs.
AI step4Lab
Choose and cross
Distinct survivors with genetic variation and heat-tolerant traits are crossed. Selective breeding, not genetic modification.
5Reef
Replant and watch
Young kelp is attached to rocky reef in pilot tests. Warm summers show which plants survive.
Sources: Google blog post, Nov 12, 2026, and Google Research video; Scharfenstein et al., BMC Genomics, 2026; CSIRO, 2024, and CSIRO on Google Arts & Culture. The scene is a schematic, not a map or real data.
Graphic: Powered by Thorsten Meyer AI
Selecting Kelp for Warmer Seas
Tasmania’s giant kelp decline puts the research in the context of a damaged marine ecosystem, not simply a new AI application. CSIRO says kelp populations have fallen by 95% from historic numbers. The agency attributes the decline to warmer, nutrient-poor water and invading sea urchins, alongside the East Australian Current moving farther south. Giant kelp forests provide habitat for marine life, including fish, octopuses and sea stars.
If the genetic work helps identify plants that can persist under warmer conditions, selective breeding could offer one tool for restoring reefs. But the reported analysis is an early step: choosing candidates does not establish that their offspring will tolerate future sea temperatures, nor that restoration will rebuild a functioning forest. Survival in the sea is the practical test, and the source material says that test has not yet taken place.
Products Worth Considering
Amazon Kindle Scribe (32GB) with Premium Pen, Tungsten (Like-New)
Bose SoundLink Max Portable Bluetooth Speaker, IP67 Waterproof and Dustproof, Up to 20 Hours of Battery Life, Black
Jackery Explorer 300 Portable Power Station, 292Wh
As an affiliate, we earn on qualifying purchases.
From Genome Maps to Wild Samples
The collaboration was announced in February 2024 by Google Australia, CSIRO and partners including the University of Tasmania’s Institute for Marine and Antarctic Studies, The Nature Conservancy, the Great Southern Reef Foundation and the Kelp Forest Alliance. The wider project had two strands: mapping kelp using satellite imagery and developing genomic information to support recovery. Google’s November update concerns the genomics strand; the original plan included using Google Earth Engine and Vertex AI to examine more than 7,000 square kilometres of imagery.
The genome study addresses a limitation in earlier comparisons. Before the Australian genomes were published, the available giant kelp reference genome came from California. The 1.5% difference reported between the Californian and Australian genomes is greater than the 0.2% difference between the two Australian examples, suggesting that a distant reference could obscure variation in local kelp. The June study is peer-reviewed and its data is public, according to the source material; this does not independently verify the later 600-sample analysis.
“trained a version of our AI model DeepVariant to analyze our wild samples”
— Google, in its video accompanying the November 12 announcement
The Field Test Is Still Ahead
The 600-sample analysis and the reported genetic isolation of remaining Tasmanian patches are currently claims in Google’s blog post and video. The source material reports no accompanying research paper or dataset for those results. It is not clear when those materials will be released or what independent review of the findings may follow.
It is also not yet clear which traits the breeding work will prioritize, how many kelp plants will be crossed, or when young kelp will be planted on reefs. Most importantly, there is no reported field evidence yet that selectively bred kelp can survive warmer summers. The published reference genomes support the underlying comparison work, but they do not by themselves prove that the restoration approach will succeed.
Breeding and Reef Trials
According to the project description, scientists plan to cross distinct native survivors in a laboratory using selective breeding, not genetic modification, then plant young kelp on reefs. The source does not give a schedule for those steps or specify a date for the first field results.
The next evidence to watch for includes publication or release of the 600-sample analysis, details of the chosen breeding candidates, and monitoring data from planted kelp. Until plants face warmer summer conditions in the sea, the central question—whether the selected kelp can persist where forests have declined—remains open.
Key Questions
What did Google and CSIRO report?
Google said researchers used a kelp-adapted version of DeepVariant to analyze 600 wild giant kelp samples from 40 sites along Tasmania’s coast. The aim is to inform selective breeding.
Has the 600-sample analysis been published?
The source material identifies the analysis and the claim that remaining patches are genetically isolated in Google’s November 12 post and video. It reports no paper or dataset for those results yet.
What has been peer-reviewed?
A study reporting two Australian giant kelp reference genomes was published in BMC Genomics in June 2026. It found a 0.2% difference between the two Australian genomes and a 1.5% difference from the Californian reference.
Is the project genetically modifying kelp?
No. The described next step is selective breeding: crossing distinct native survivors in a laboratory, then planting young kelp on reefs.
Has the bred kelp been shown to survive warmer summers?
No field-survival result is reported. Whether the selected kelp can survive warmer summers in the sea remains the project’s key unanswered test.
Source: ThorstenMeyerAI.com
