Tom Hope

Scimon🧪: Scientific Inspiration Machines Optimized for Novelty. LLMs + inspiration retrieval from papers + 🚀iterative novelty boosting🚀: 1. LLM (e.g. GPT4) generates idea 2. Compares to detected prior work 3. Refines idea novelty 4. Repeat until sufficient novelty reached

I am hiring a Research Scientist at the University of Arizona in AI/NLP to work on automated scientific feasibility assessment for scientific discovery, particularly through automated experiments. Remote work within the US can be considered. Application: arizona.csod.com/ux/ats/careers…

I am actively seeking a PhD summer intern at @allen_ai to focus on test-time learning (/other learning approaches) to improve the generation of {novel, surprising, high-utility, etc.} hypotheses in the context of scientific discovery. Please DM/email by 3/18 if interested. A brief pitch when contacting would be wonderful. I may not be able to reply to all, but if it aligns, I'll be sure to contact you. Prior experience with TTL, representation learning, and model editing would be preferred. Also, please apply with the link in the thread if you haven't already.

Are you a researcher in CS or a CS-adjacent field curious about how an AI agent can help you with your research project? Want to try a new tool for your research support in a paid user study ($100, 2 hr)? Limited spot numbers. See details and sign up here: forms.gle/JzLtkAhe7Ttvui…

Enjoyed giving this talk on AI for recombining ideas across research papers to generate novel directions, and AI for assessing and optimizing novelty. Lots of work ahead for building creative AI and assessing its creativity!

@percyliang @percyliang you may find our work at @allen_ai on forecasting and simulating scientific trajectories interesting. Simulating a “society of scientists”, in a way. Led by the excellent @anirudhajith42 .

I think it’s pretty clear that simulation is the next frontier for AI. The most impressive feats of AI to date are when we have a clear environment + reward, whether it be beating Le Sedol at Go, winning an IMO gold medal, or writing entire apps from scratch. In these cases, the RL algorithm can try different actions, and observe the well-defined consequences in the safety of a docker container. But what about messy real-world situations involving people? The rewards are unclear, the stakes are high, and you can’t experiment in the real world. But these situations are precisely where the next big opportunity in AI is. To crack this, we need to *simulate* society (“put society into a docker container”). Concretely, this means building a model that can predict what will happen in any given situation (real or hypothetical). If we can do this, we are only limited by our imagination: predict the future, optimize for better outcomes, answer hypothetical (“what if”) questions. Ultimately, this goes beyond making better decisions, but it’s about giving us a better understanding of ourselves and the world. Simulation is the whole enchilada. And this is exactly the research that @simile_ai is working on. Read more here: simile.ai/blog/simulatio…

Given the record of scientific research so far, can we predict what comes next? I learned a lot working on PreScience with @_DougDowney @Hoper_Tom @aps6992 @jaydepun and our friends from @UChicago and @allen_ai! Check out our paper and try your hand at our benchmark!

@anirudhajith42 @_DougDowney @aps6992 @jaydepun @UChicago @allen_ai It’s been great working with you !

Can today’s agents anticipate future scientific collaborations, ideas, and impact? Introducing PreScience, a large-scale AI benchmark for scientific forecasting. Careful dataset construction led by @anirudhajith42, with @aps6992, @jaydepun, @Hoper_Tom and collaborators.

Check out our new paper on scientific forecasting. We predict which research teams will assemble, what prior work they will build upon together, what new scientific advances they will make (ideas and findings), and what the downstream citation impact will be.

New preprint 💥 Can a general-purpose model achieve results comparable to medically pre-trained models? 🤔 We show that lightweight fine-tuning of a general-purpose LVLM and an LVLM-aware retriever can. 🚀 🔗 GitHub: github.com/Nirmaz/CLARE 📄 Paper: arxiv.org/pdf/2508.17394

Check out our new work on predictive multimodal RAG for the medical domain. Retrieving medical images from literature and hospital records in a way that is directly tuned for prediction.

🎉 Proud to share that "Debatable Intelligence" has now been accepted to #EMNLP2025 (Main Conference)! noy-sternlicht.github.io/Debatable-Inte… Huge thenks to my amazing collaborators @ArielGera2, @RoyBarHaim, @Hoper_Tom, @noamslonim

For those interested in LLMs for Scientific Discovery: Sign up to be a potential reviewer for this new proposed workshop at *ACL!

We are proposing the second workshop on AI & Scientific Discovery at EACL/ACL. The workshop will explore how AI can advance scientific discovery. Please use this Google form to indicate your interest: docs.google.com/forms/d/e/1FAI… More in the 🧵! Please share! #MLSky 🧠

1/ 📢 @CosmikNetwork has been awarded $1M in grant funding by @open_phil and @AsteraInstitute! These generous grants will support our development of Semble - a social "micro-knowledge" network for researchers. Think Are.na + Goodreads for research!

Paper: arxiv.org/abs/2506.22026 @its_sshahid @marissa_rad @rayrayfok @Siangliulue @dsweld @allen_ai

Best paper award at @sdpworkshop @aclmeeting : Our paper on evaluating the novelty of LLM-generated scientific ideas won the best paper award at the SDP workshop. Congrats Simra @its_sshahid Marissa @marissa_rad and team!

🔔 New Paper! We propose a challenging new benchmark for LLM judges: Evaluating debate speeches. Are they comparable to humans? Well... it’s debatable. 🤔 noy-sternlicht.github.io/Debatable-Inte… 👇 Here are our findings:

7/7 Led with the guidance of @Hoper_Tom — thank you! Paper - arxiv.org/abs/2505.20779 Project page - noy-sternlicht.github.io/CHIMERA-Web Github (code + data) - github.com/noy-sternlicht…