🤖🔬Today we are debuting Zochi, the world’s first Artificial Scientist with state-of-the-art contributions accepted in ICLR 2025 workshops. Unlike existing systems, Zochi autonomously tackles some of the most challenging problems in AI, producing novel contributions in days—from idea to finalized publication. With a standardized automated reviewer, Zochi’s papers score an average of 7.67 compared to other publicly available papers generated by AI systems that score between 3 and 4. 🧵👇
Zochi’s work covers diverse and open-ended problems at the frontier of AI Read our full technical report here → https://t.co/Bn3jyVcsMr 1️⃣ CS-ReFT finetunes LLMs at the subspace level, enabling the much smaller Llama-2-7B to surpass GPT-3.5's performance using only 0.0098% of
Zochi’s papers received unanimously positive reviews: 1️⃣ CS-ReFT received strong peer review scores (6,7,6), with reviewers commending its "clever idea" and effectiveness in addressing "a critical limitation of ReFT." 2️⃣ Reviewers gave Siege scores of (7,7), highlighting the
In an exploratory exercise on MLE-Bench Kaggle competitions, Zochi achieves the state-of-the-art without any task-specific optimization, surpassing median human performance on 80% of tasks and securing medals on 50% of them.
We're committed to maintaining scientific integrity. Since AI systems cannot take responsibility for their work, we do not believe they should be listed as authors and discourage the submission of AI-produced work without thorough human verification. We are in discussion with the
At Intology, we're building Artificial Scientists because we believe they are the ultimate use case of artificial intelligence. By accelerating the pace of discovery across all fields, we can help address humanity's most significant challenges. Learn more: