Adaption aims big with AutoScientist, an AI tool that helps models train themselves

Adaption aims big with AutoScientist, an AI tool that helps models train themselves

In an exciting development for the artificial intelligence landscape, Adaption has unveiled AutoScientist, an innovative tool designed to enhance the self-training capabilities of AI models. After years of anticipation from researchers, this launch signifies a significant leap towards achieving AI systems that can improve themselves more effectively than human intervention. This new offering comes at a time when investment in research-driven AI labs is at an all-time high, providing the necessary resources to turn ambitious ideas into reality. AutoScientist employs an automated fine-tuning process, enabling models to learn specific skills more efficiently. While the technology can be applied across various domains, the team at Adaption is particularly focused on its potential to streamline the training and fine-tuning of advanced AI models. Sara Hooker, co-founder and CEO of Adaption and former VP of AI research at Cohere, highlighted the transformative nature of AutoScientist. According to her, the tool optimizes both the data and the model simultaneously, paving the way for successful training of frontier AI outside traditional lab environments. "What’s super exciting about it is that it co-optimizes both the data and the model, and learns the best way to basically learn any capability," Hooker explained in an interview. Building on Adaption's existing product, Adaptive Data, which simplifies the creation of high-quality datasets, AutoScientist aims to convert these continuously evolving datasets into equally advanced AI models. Hooker emphasizes that the entire stack should be adaptable and capable of optimizing in real-time for any task. Adaption claims that AutoScientist has significantly increased win rates across multiple models, though the exact impact of these figures remains to be fully contextualized. As the system is tailored for specific tasks, traditional benchmarks may not apply. Nevertheless, the company is optimistic that users will recognize the improvements firsthand. To encourage exploration of this new tool, Adaption is offering a free 30-day trial upon its release. "Just as code generation has unlocked numerous tasks, this innovation will catalyze advancements across various fields," Hooker added.

Sources : TechCrunch

Published On : May 13, 2026, 12:25

Cybersecurity
The Elusive Phineas Fisher: The Hacktivist Who Took Down Spyware Giants

In the realm of cybersecurity, few figures are as intriguing as Phineas Fisher, a hacker who has evaded capture for near...

TechCrunch | Jul 25, 2026, 21:00
The Elusive Phineas Fisher: The Hacktivist Who Took Down Spyware Giants
Streaming
Kalshi Challenges Netflix Over Controversial Documentary Trailer

Kalshi, the prediction market platform, has taken significant legal steps against Netflix, sending a cease-and-desist le...

TechCrunch | Jul 25, 2026, 17:10
Kalshi Challenges Netflix Over Controversial Documentary Trailer
Startups
The Boring Company Eyes $4 Billion Funding Boost Amid Expanding Tunnel Ventures

Elon Musk's tunneling enterprise, The Boring Company, is reportedly negotiating a substantial funding round of $4 billio...

TechCrunch | Jul 25, 2026, 19:50
The Boring Company Eyes $4 Billion Funding Boost Amid Expanding Tunnel Ventures
AI
The Shift in Human Cognition: Embracing AI as a Collaborative Tool

As technology continues to evolve, a notable shift is occurring in the relationship between humans and artificial intell...

Business Insider | Jul 25, 2026, 09:50
The Shift in Human Cognition: Embracing AI as a Collaborative Tool
Startups
AI Transformations Lead to Major Job Cuts at Tech Giants

Monday.com, the innovative work management platform based in Tel Aviv, has recently announced significant layoffs, attri...

TechCrunch | Jul 26, 2026, 01:45
AI Transformations Lead to Major Job Cuts at Tech Giants
View All News