using AI
https://www.science.org/content/article/problems-22-scientific-fields-had-solutions-hiding-plain-sight-ai-has-found-them
https://www.anthropic.com/research/claude-shaped-science
A researcher is using AI to solve problems in multiple fields of study. He is basically using methodology that works in one field to solve problems in other fields.-a I tried to get an actual reprint of one of the 30 papers from 22 scientific fields, but they do not seem to be available.-a The link given takes you to a blog that talks about the achievements, but has no drafts of the research papers.-a The one that I wanted to look at was the Human, chimp, gorilla, and orang phylogeny paper where they found 2 phylogenies had an equal probablity of being inferred from the data.-a The blog added additional information, and it was only dealing with single gene phylogenies.-a No one uses single gene analysis anymore in this age of genomics, and we never relied on it even for the first ape phylogenies that utilized three or four ape genes to resolve the relationship between humans, chimps and gorillas.-a Single genes can give you unexpected results.-a There are a lot of factors and it
isn't just random noise, but you can have horizontal gene transfer, and you can have gene conversion from dupicated genes within the genome that would alter the evolutionary history of that sequence.-a What the AI did was identify the genes that did not give statistically significant results between possible phylogenies.-a These are the genes that you would not use in any analysis except if they were the majority of the genes tested.-a If that was the case then the genes could be telling a consistent story about ambiguities that might exist in certain portions of the phylogeny.
Ron Okimoto
On 2026/10/11 7:12 PM, RonO wrote:
https://www.science.org/content/article/problems-22-scientific-fields-had-solutions-hiding-plain-sight-ai-has-found-them
https://www.anthropic.com/research/claude-shaped-science
A researcher is using AI to solve problems in multiple fields of study. He is basically using methodology that works in one field to solve problems in other fields.a I tried to get an actual reprint of one of the 30 papers from 22 scientific fields, but they do not seem to be available.a The link given takes you to a blog that talks about the achievements, but has no drafts of the research papers.a The one that I wanted to look at was the Human, chimp, gorilla, and orang phylogeny paper where they found 2 phylogenies had an equal probablity of being inferred from the data.a The blog added additional information, and it was only dealing with single gene phylogenies.a No one uses single gene analysis anymore in this age of genomics, and we never relied on it even for the first ape phylogenies that utilized three or four ape genes to resolve the relationship between humans, chimps and gorillas.a Single genes can give you unexpected results.a There are a lot of factors and it
isn't just random noise, but you can have horizontal gene transfer, and you can have gene conversion from dupicated genes within the genome that would alter the evolutionary history of that sequence.a What the AI did was identify the genes that did not give statistically significant results between possible phylogenies.a These are the genes that you would not use in any analysis except if they were the majority of the genes tested.a If that was the case then the genes could be telling a consistent story about ambiguities that might exist in certain portions of the phylogeny.
Ron Okimoto
Anthropic's advertising does not match reality. The "papers" appear to have not been published anywhere that I can discover. Dr. Sabine Hossenfelder also tried to find the papers, and she failed.
| Sysop: | Amessyroom |
|---|---|
| Location: | Fayetteville, NC |
| Users: | 76 |
| Nodes: | 6 (0 / 6) |
| Uptime: | 59:25:14 |
| Calls: | 1,265 |
| Calls today: | 1 |
| Files: | 1,362 |
| D/L today: |
4 files (9,632K bytes) |
| Messages: | 286,017 |