Resources presented in the discussion

The discussion provided an overview of

  • some new helpful AI tools for biomedical research
  • examples of success stories using AI effectively for research
  • stories of caution or failure identifying weaknesses or areas where researchers should be especially careful
New Helpful AI Tools

Molmo is an AI tool for understanding and interacting with visual data

AlphaFold Server to predict structure of various biological complexes

AlphaGenome to predict gene expression levels and the impact of mutations

There are several tooling platforms for sharing code, data, and benchmarks

Success Stories

The Virtual Lab of AI agents designs new SARS-CoV-2 nanobodies

CRISPR-GPT for agenic automation of gene-editing experiments

Prompt Adaptation as a Dynamic Complement in Generative AI Systems

Measuring AI Ability to Complete Long Tasks

Failures

Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer Productivity

I destroyed months of work by a developer in seconds says AI coding tool after panicking and deleting an entire database

Does ChatGPT Ignore Article Retractions and Other Reliability Concerns?

Multi-model assurance analysis showing large language models are highly vulnerable to adversarial hallucination attacks during clinical decision support

Emergent Misalignment: Narrow finetuning can produce broadly misaligned LLMs

Deep-learning based gene perturbation effect prediction does not yet outperform simple linear baselines

Expressing stigma and inappropriate responses prevents LLMs from safely replacing mental health providers

Endoscopist deskilling risk after exposure to artificial intelligence in colonoscopy: a multi-center, observational study

On Targeted Manipulation and Deception when Optimizing LLMs for User Feedback

Suggested Optional Activity

Because this workshop is discussion based, there is no guided activity. However, we provide an area with some further resources. Specifically this area regards prompt engineering and how to write effective prompts when interacting with an AI tool. We saw the positive and negative effect of prompts in several of the discussion examples and view this as an important area for developers and researchers to consider more deeply.