AGENTIC-SYSTEMSMay 16, 20261 min read8 views

Uniting biological toolkits for a new approach to ALS

By leveraging the AI tool Co-Scientist, researchers Ritu Raman and Ryan Flynn have combined their distinct expertise in tissue engineering and RNA biology to accelerate the search for novel ALS treatments.

Ritu Raman at MIT and Ryan Flynn at Boston Children's Hospital approach human biology with very different toolkits, but Co-Scientist is bridging their labs. Raman, a mechanical engineer, builds living nerve and muscle tissues to model diseases that affect voluntary movement. Her husband Flynn, a chemical biologist, maps RNA on the surface of cells to see how it influences cellular communication and how pathogens invade.

When Raman decided to investigate ALS, which was outside of her usual domain, she faced a sprawling, contradictory literature that would usually take months to grasp. Co-Scientist compressed that work, quickly helping Raman interrogate the evidence in relation to her tissue model, turn ideas into testable hypotheses, and rank potential directions in accordance with the trade-offs labs actually face, such as feasibility and potential risk–reward.

But Co-Scientist’s best leads came with a catch: they involved what happens at the surface of cells, where much of their communication is mediated. Raman could manipulate tissues and measure outcomes, but decoding the molecular interactions driving those signals was outside her area of expertise.

That gap became the catalyst for collaboration. Raman brought her new research directions to Flynn, and the pair used Co-Scientist iteratively, combining its best ideas into creative research pathways that united their distinct toolkits. To develop new therapies, their hunt is now on for novel RNA-based mechanisms—and potentially RNA-based drugs—that could be used to target ALS.

Source: Google DeepMind Blog

More in this category

agentic-systems

Generative Ontology Induction: Domain-Agnostic Schema Discovery from Document Corpora Using Large Language Models

Researchers introduce Generative Ontology Induction (GOI), a domain-agnostic framework that automatically extracts structured ontologies from document corpora using LLMs. Achieving 95-100% structural coverage, GOI addresses a major bottleneck in knowledge-intensive AI systems.

agentic-systems

JUMP: Single-Pass Membership Inference on Fine-Tuned Diffusion Language Models

Researchers have proposed JUMP, a novel single-pass membership inference attack designed for fine-tuned discrete diffusion language models (dLLMs). By leveraging the unique properties of dLLMs, JUMP significantly improves detection accuracy while drastically reducing the number of required queries.

agentic-systems

Masked Diffusion Language Models are Strong and Steerable Text-Based World Models for Agentic RL

Researchers demonstrate that Masked Diffusion Language Models (MDLMs) serve as highly effective, steerable text-based world models for agentic reinforcement learning. By leveraging bidirectional denoising, MDLMs outperform autoregressive models four times their size in coherence, groundedness, and rollout diversity.

NOW LET US Related – Democratizing AI with Small Language Models: Structured Benchmarking and Parameter-Efficient Fine-Tuning for Local Deployment

agentic-systems

Democratizing AI with Small Language Models: Structured Benchmarking and Parameter-Efficient Fine-Tuning for Local Deployment

A new study demonstrates that small language models (SLMs) under 3 billion parameters can serve as highly capable local experts for specialized tasks. By combining structured benchmarking with low-cost parameter-efficient fine-tuning (PEFT), institutions can achieve AI autonomy without relying on expensive hardware.

agentic-systems

PlanFlip: Attacking Multi-Agent LLM Systems via Planning-Phase Prompt Injection

Researchers have introduced PlanFlip, a novel prompt injection attack framework targeting the planning phase of multi-agent LLM systems. The study reveals critical security blind spots in homogeneous agent pipelines and demonstrates that reasoning-augmented models like DeepSeek-R1 exhibit strong resistance.

agentic-systems

Rater State Bias in RLHF Preference Data: An Audit Framework

A new study identifies a structured bias in Reinforcement Learning from Human Feedback (RLHF) caused by the psychological state of human raters. Under stress, raters' shifting preferences can propagate through reward modeling, potentially compromising AI policy optimization.