By using this site, you agree to the Privacy Policy and Terms of Use.
Accept
Online Tech Guru
  • News
  • PC/Windows
  • Mobile
  • Apps
  • Gadgets
  • More
    • Gaming
    • Accessories
    • Editor’s Choice
    • Press Release
Reading: A New Kind of AI Model Lets Data Owners Take Control
Best Deal
Font ResizerAa
Online Tech GuruOnline Tech Guru
  • News
  • Mobile
  • PC/Windows
  • Gaming
  • Apps
  • Gadgets
  • Accessories
Search
  • News
  • PC/Windows
  • Mobile
  • Apps
  • Gadgets
  • More
    • Gaming
    • Accessories
    • Editor’s Choice
    • Press Release

The Best Prime Day Laptop Deals on Computers That Are Actually Worth Buying

News Room News Room 10 July 2025
FacebookLike
InstagramFollow
YoutubeSubscribe
TiktokFollow
  • Subscribe
  • Privacy Policy
  • Contact
  • Terms of Use
© Foxiz News Network. Ruby Design Company. All Rights Reserved.
Online Tech Guru > News > A New Kind of AI Model Lets Data Owners Take Control
News

A New Kind of AI Model Lets Data Owners Take Control

News Room
Last updated: 9 July 2025 19:22
By News Room 4 Min Read
Share
SHARE

A new kind of large language model, developed by researchers at the Allen Institute for AI (Ai2), makes it possible to control how training data is used even after a model has been built.

The new model, called FlexOlmo, could challenge the current industry paradigm of big artificial intelligence companies slurping up data from the web, books, and other sources—often with little regard for ownership—and then owning the resulting models entirely. Once data is baked into an AI model today, extracting it from that model is a bit like trying to recover the eggs from a finished cake.

“Conventionally, your data is either in or out,” says Ali Farhadi, CEO of Ai2, based in Seattle, Washington. “Once I train on that data, you lose control. And you have no way out, unless you force me to go through another multi-million-dollar round of training.”

Ai2’s avant-garde approach divides up training so that data owners can exert control. Those who want to contribute data to a FlexOlmo model can do so by first copying a publicly shared model known as the “anchor.” They then train a second model using their own data, combine the result with the anchor model, and contribute the result back to whoever is building the third and final model.

Contributing in this way means that the data itself never has to be handed over. And because of how the data owner’s model is merged with the final one, it is possible to extract the data later on. A magazine publisher might, for instance, contribute text from its archive of articles to a model but later remove the sub-model trained on that data if there is a legal dispute or if the company objects to how a model is being used.

“The training is completely asynchronous,” says Sewon Min, a research scientist at Ai2 who led the technical work. “Data owners do not have to coordinate, and the training can be done completely independently.”

The FlexOlmo model architecture is what’s known as a “mixture of experts,” a popular design that is normally used to simultaneously combine several sub-models into a bigger, more capable one. A key innovation from Ai2 is a way of merging sub-models that were trained independently. This is achieved using a new scheme for representing the values in a model so that its abilities can be merged with others when the final combined model is run.

To test the approach, the FlexOlmo researchers created a dataset they call Flexmix from proprietary sources including books and websites. They used the FlexOlmo design to build a model with 37 billion parameters, about a tenth of the size of the largest open source model from Meta. They then compared their model to several others. They found that it outperformed any individual model on all tasks and also scored 10 percent better at common benchmarks than two other approaches for merging independently trained models.

The result is a way to have your cake—and get your eggs back, too. “You could just opt out of the system without any major damage and inference time,” Farhadi says. “It’s a whole new way of thinking about how to train these models.”

Share This Article
Facebook Twitter Copy Link
Leave a comment

Leave a Reply Cancel reply

Your email address will not be published. Required fields are marked *

Astro Bot Finally Gets 2 Long-Awaited, Fan-Requested Bots

News Room News Room 10 July 2025
FacebookLike
InstagramFollow
YoutubeSubscribe
TiktokFollow

Trending

Amazon Prime Day 2025 Sale: Early Deals on OnePlus Nord 5, iQOO Z10 5G, Samsung Galaxy M16 5G and More

The Amazon Prime Day 2025 sale gets underway on July 12. The annual sale is…

10 July 2025

The ModRetro Chromatic Is a Game Boy Fit for Your Apocalypse Bunker

For that reason, the console exclusively plays cartridge games, just like its progenitor. There are…

10 July 2025

The Verge’s favorite desk accessories for Prime Day 2025

What you have on your desk and in your workspace at home can reveal a…

10 July 2025
News

I never leave home without Anker’s power bank that has built-in cables

Anker’s 25,000mAh power bank with two built-in, retractable USB-C cables might be as essential for you as it is for me. It offers fast charging speeds, serving up to 100W…

News Room 10 July 2025

Your may also like!

Accessories

Indian Startup QWR Unveils AI-Powered Smart Glasses Humbl, to Be Shipped Later This Year

News Room 10 July 2025
PC/Windows

Acer Iconia Tab iM11 Price, Specifications, Features, Comparison

News Room 10 July 2025
Mobile

Samsung to Launch Its First Tri-Fold Smartphone by the End of This Year: Report

News Room 10 July 2025
News

Elon Musk Says Grok Is Coming to Tesla EVs

News Room 10 July 2025

Our website stores cookies on your computer. They allow us to remember you and help personalize your experience with our site.

Read our privacy policy for more information.

Quick Links

  • Subscribe
  • Privacy Policy
  • Contact
  • Terms of Use
Advertise with us

Socials

Follow US
Welcome Back!

Sign in to your account

Lost your password?