Jon Krohn

🚀 Transforming Grant Writing with AI and Behavioral Science We’re excited to share our latest work: “Implementation Plan for the Retrieval-Augmented Generation System in the GRANT-AI Trial.” This paper details how the GRANT-AI pilot integrates Retrieval-Augmented Generation (RAG) and expert human review to revolutionize NIH grant proposal development. By combining GPT-4o, a curated corpus of funded proposals, and behavioral design elements like motivational nudges and progress visualizations, GRANT-AI offers a scalable, ethical, and evidence-based approach to improving proposal quality and equity in research funding. Key innovations include: 🧠 Integration of expert human reviewers and adaptive AI feedback. 💡 Behavioral scaffolding to enhance writing motivation and completion. 🔍 A secure, NIH-aligned RAG feedback engine built for accuracy and contextual relevance. Our goal? To make grant writing more efficient, equitable, and data-driven for researchers everywhere. 📄 Read the full implementation plan here: https://lnkd.in/d69svhg5 and join the conversation on how AI can responsibly augment scientific writing and peer review. #SporeDataResearch #AIinScience #GrantWriting #RAG #OpenAI #BehavioralScience #ResearchInnovation #NIH 

1

Stop Sounding Like An AI: The Voice DNA Method You've felt it. That wrongness when Claude writes for you. Structure's solid, facts are right, but it doesn't sound like you. It sounds like an AI wrote it. Fix it in 10 minutes. No repeating yourself every conversation. What is Voice DNA? A config file that tells Claude how you write: • Writing rules — sentence structure, hedging, specificity • Banned phrases — "delve," "leverage," "game-changer" • Your writing samples — so Claude pattern-matches your voice The Problem Default Claude: "In today's rapidly evolving AI landscape, it's worth noting that leveraging these cutting-edge tools can help you unlock significant productivity gains." Voice DNA Claude: "AI tools save time if you configure them right." Same thing. One sounds human. Three Ways to Apply 1. User Preferences (Recommended) Settings → Profile → User preferences → Paste content → Save Works globally across all conversations. 2. Project Custom Instructions Add to Claude Project knowledge → Set custom instructions Project-specific voice control. 3. Per-Conversation Paste at conversation start One-off use only. What's Inside Writing Guidelines: • Get to the point • Vary sentence length • Hedge when uncertain • Be specific • Never pad output Banned Phrases: • Delve / Dive into / Unpack • Harness / Leverage / Utilize • Game-changer / Cutting-edge • "This isn't X. This is Y." (FATAL) Works immediately. Add your writing samples to match your specific voice. Making It Yours Add 3-5 writing samples: • Old emails (thoughtful ones) • Google Docs • Blog posts • Slack messages • Reports Critical: use pre-2023 writing. Recent stuff already has AI fingerprints. How to Verify Start new conversation. Ask Claude to write in your domain. Check for: • Shorter sentences • No banned phrases • Less hedging • Your tone Why It Works Language models learn patterns. Give Claude your rules and samples, it adjusts probability distribution toward your style. Banned phrases eliminate tells. Writing rules shift structure. Samples provide targets. Result: passes the "sounds like me" test. Bottom Line Set it once in User Preferences. Forget about it. Ten minutes now. Consistent voice forever. Get the template: https://72ai.in/voice-dna/

5

The "New Renaissance": Demis Hassabis on How AI Will Solve the Hardest Problems in Science and Life. A must watch for those of you who havent seen this interview. In a wide-ranging interview with Fortune, Google DeepMind CEO Demis Hassabis outlines a future where AI moves beyond chatbots to become a fundamental tool for scientific discovery and a universal assistant for daily life. The Scope & Future of AI: Hassabis predicts that if we deploy AI correctly, we are entering a "new golden era of discovery" comparable to the Renaissance. He envisions a world of "radical abundance" within 10-15 years, driven by AI’s ability to accelerate solutions in energy, material science, and travel. Key Use Cases & Benefits Revolutionizing Drug Discovery (In Silico Biology): Moving beyond just identifying protein structures (AlphaFold), the next phase is designing drugs entirely in simulation (in silico). Hassabis notes this approach can cut the costs and time of drug development by orders of magnitude, turning a 10-year process into something much faster. His new venture, Isomorphic Labs, is already in pre-clinical trials for cancer and cardiovascular drugs. The Rise of "Agents" (Action over Talk): We are crossing the watershed moment where AI models stop being just chatbots and become "Agents"—capable of planning, reasoning, and executing tasks. This shift will happen in earnest by late 2026. The "Universal Assistant": Hassabis describes a future assistant that isn't trapped in one app but works across your phone, browser, and even smart glasses. It will understand your context implicitly - whether you are coding, handling admin, or navigating the real world - and offer help proactively. Solving Hard Science: Beyond biology, AI will be crucial for material science breakthroughs needed for fusion energy, optimal batteries, and solar power efficiency. Quote of the interview: "I think human health will be revolutionized... we'll have solved many big problems in science... and I think we'll be in a world of radical abundance." - Demis Hassabis Watch the full interview: https://lnkd.in/gFT5PZnh #AI #ArtificialIntelligence #DeepTech #LifeSciences #FutureOfHealth #Innovation #ScientificDiscovery

5

If you're trying out OpenAI's o3-Pro this morning and looking for some guidance in how to use or leverage it, this is for you. https://lnkd.in/gE2B7PbC

3

Andrew Goldman, EVP, HMH Labs | 2026 prediction: “The focus of AI in education will shift from outputs to outcomes.” In this clip from our year-end predictions series, Andrew shares a hopeful but important prediction about where AI needs to go next. AI adoption is accelerating because it genuinely helps educators who are stretched across too many demands. It solves real, immediate pain points. But Andrew argues that the next phase can’t stop at productivity or outputs alone. In 2026, the real question will be how AI supports learning outcomes in meaningful, measurable ways, not just whether it makes tasks faster or easier. Listen to more predictions from Andrew and other EdTech leaders in Year-End Special Part 1: Reviewing 2025 + 2026 AI & EdTech Predictions 👉 https://lnkd.in/e2eZUxGM #EdTech #AIinEducation #LearningOutcomes #FutureOfLearning

10

1 Comment

The latest UPCEA report, The Future is Now: Essential Conversations for Building Tomorrow’s University Today, captures exactly where higher education is heading — and why our partnerships at Riipen are focused on helping institutions get there faster. UPCEA highlights a few urgent truths: - Students are making choices based on ROI and career outcomes - Credentials need to be stackable and connected to real career progress - Institutions must strengthen collaboration with employers to keep programs relevant - Professional, Continuing, and Online (PCO) units will lead the way in innovation At Riipen, we see this shift every day in our partnerships. Together with universities and colleges, we’re building industry micro-credentials combined with project-based work experiences - helping learners demonstrate skills, helping educators make curriculum more applied, and helping employers find talent ready to contribute on day one. By connecting industry, education, and learners through authentic projects, we’re helping institutions turn these UPCEA insights into scalable, measurable action. You can read the full UPCEA report here: https://lnkd.in/gxw7iSh2

41

3 Comments

OpenAI has teamed up with Retro Biosciences to create GPT-4b micro, a biological AI model trained specifically on biology and protein engineering data. https://buff.ly/8iwZRZD

2

Grant funders are bombarded by funding applications. Can Generative AI help with grant peer review? One funder in Barcelona (Fundación ”la Caixa”) has already implemented it to assist with peer review of grant proposals. Now, the UK Research and Innovation is opening up between 1,000-2,000 of its past grant applications to test out if it can/should make use of generative AI. My latest for Chemistry World: https://lnkd.in/gdQ6_Qit Mike Thelwall, Mohammad Hosseini

14

This viral post about the pace and capabilities of AI is arguably a little hyperbolic, but I honestly don’t believe by much. Based on my own recent interactions with Claude’s latest model, and Claude Code, a lot of what he is saying absolutely resonates with me. https://lnkd.in/gyW9NtjU

11

4 Comments

ICYMI 🎧 Robot dogs as tutors. AI in higher ed. Scaling Socratic learning. Stanford bioengineer Jan Liphardt lays out a bold new vision of education—where multi-agent systems help humans think better together. Catch up on The Learning Hack. 🎙️ LH115: Meet Your New Math(s) Teacher: a Robot Dog Watch/listen now: links in comments Thanks to our sponsors World of Learning Synthesia iVentiv #EdTech #LearningHack #JanLiphardt #OpenMind

7

1 Comment