đ§âđŹâđ¤ Agentic Decentralized GeroScience solves the Longevity Problem https://nitter.gq/t.co/9PlIAjEAwu đˇ Hard Hat Areađ§
USA
Joined April 2025
- Tweets7K
- Following2.1K
- Followers342
- Likes795
AI DeSci Guy retweeted
Few people have shaped #DrugDiscovery like Greg Verdine. The Harvard professor emeritus and DoveTree founder helped drug the cancer targets RAS(ON) and β-catenin, long considered âundruggable,â and founded more than a dozen biotech companies. We recently welcomed him to XtalPi.
AI DeSci Guy retweeted
New Nvidia + other labs paper shows that irrelevant text hurts mainly by weakening the model's use of the question, not by directly pushing against the right answer.
Prompts in RAG, tool use, and long-context apps often carry text that has nothing to do with the question. Even a line like "Computers are useful" lowers a model's support for the right answer.
They traced this inside open models and found intrusion heads, a small set of attention heads that read the unrelated text and weaken the answer. Masking those heads' attention to the context leaves the heads that use helpful evidence untouched.
Across 4 tasks on 7B-scale models, masking about 10 heads recovered 57% of the accuracy lost to distractors on average. An oracle fix that needs a matching clean run recovered 97.6%.
If you serve open-weight models with noisy retrieved context, a training-free mask on a few model-specific heads can win back much of the lost accuracy.
AI DeSci Guy retweeted
Some takeaways after a few months of building an automated lab for tissue manufacturing:
- Software and AI can do a lot of the troubleshooting for you and integrate instruments in ways that werenât possible before. For advanced OOD workflows, though, the bottleneck quickly becomes hardware (temperature control, dead volumes, latency, etc.).
- As more powerful non-destructive measurement modalities come online, workflows need to accommodate them. Designing and planning for lean processes becomes critical.
- The teams building truly novel simulations cannot be far from the data generation engine. If you come visit us, youâll find our AI scientists in the lab talking to our tissue engineers.
(P.S. come see some tissues! It is fun!)
AI DeSci Guy retweeted
Paralyzed man walking....brainâspine interface in a man paralyzed by spinal cord injury.
AI decoded his intention to walk and stimulated his spinal cord below the injury.
He regained the ability to stand and walk in everyday settings.
doi.org/10.1038/s41586-023-0âŚ
AI DeSci Guy retweeted
Today we're launching Ordo Labs (YC F26) đ
Every medicine you take is tested in a regulated lab, where scientists still analyze data, check QC and write reports by hand.
Ordo does that work for them. In paid pilots, labs cut post-analysis time by 90%+.
Built by @SebApelgren, @kristinahoang_ and @adamtovatt
AI DeSci Guy retweeted
Lots of cancer patients and their caregivers are turning to AI to build tools and services to help track treatments and explore options that their care teams may not have suggested - I wrote about these people going "founder mode" on their cancer for The American Journal of Managed Care: bit.ly/4B4Z4Ai
âBoth of my parents each got diagnosed with cancer within a week of each other.â
Ashley is the VP of Product at @OpenAI for Health and this is a very special episode for me. We talked about how we both use AI to help our parents fight cancer â including preparing for doctor visits, researching treatments, and coping with our emotions as caregivers. Topics we discussed:
â Do doctors mind when patients bring AI research?
â How ChatGPT caught a risky treatment before it was given
â How to make time for your parents with a demanding job
Some quotes from Ashley:
âI always felt like there was no more time in the day. Then someone gets sick and you realize what you can make time for and what can drop.â
âYou also have to care for yourself as a caregiver. You have to make sure youâre budgeting time for your own mental health.â
âThe treatment she was about to be given would have had a really bad contraindication. She narrowly avoided it because she ran it through Chat.â
đ Watch now: youtu.be/pR7NCYKsA3o
Thanks to our sponsors:
@meetgranola: AI meeting notes that donât suck granola.ai/peter
@RiversidedotFM: All-in-one AI studio for podcasts and video creators.riverside.com/PeterâŚ
AI DeSci Guy retweeted
NYC With a Muslim Mayor
#ZohranMamdani
Who hired illegals Radical Islamic âŞď¸ acting as police officers?!!!
WTF?
Fake NYC đşđ¸ cops arrested by ice
hours after situation
with woman and her dog đś đđ¤ˇââď¸
AI DeSci Guy retweeted
đ¨: Cannabinoids can trigger cancer cell death and block tumor growth.
Emerging research into the endocannabinoid system suggests that compounds like THC and CBD may possess significant anti-tumor properties.
AI DeSci Guy retweeted
Longevity Grok, a.k.a. your personal longevity superintelligence (PLSI), is an open-source multi-agent workhorse designed to keep you alive & healthy for as long as possible. A collaborative effort from @InSilicoMeds & the Sinclair Lab @harvardmed. Visit longevitygrok.si
AI DeSci Guy retweeted
The broccoli derivative sulforaphane extends lifespan by slowing the transcriptional aging clock
biorxiv.org/content/10.1101/âŚ
AI DeSci Guy retweeted
NajwaĹźniejsze wydarzenia ze Ĺwiata sztucznej inteligencji z ostatnich godzin.
WspĂłĹtwĂłrca OpenAI zapowiada pokonanie biologicznej Ĺmierci, Google Ĺamie klucze RSA, nowy model czyta 100 milionĂłw tokenĂłw na raz, a rozwĂłj AI jest szybszy niĹź zakĹadaliĹmy.
Oto zestawienie najciekawszych tematĂłw dnia:
AI DeSci Guy retweeted
Lifeâs short. Letâs fix that.
There is no application of AI and SI more impactful than Productive Longevity. Personal Longevity Superintelligence brought to you by my team and brilliant @davidasinclair and made possible with @bot . Grok Bot users get longevity superpowers - link at LongevityGrok.si
AI DeSci Guy retweeted
New on HCN: the midterms and what they could mean for healthcare, from ACA subsidies to Medicaid to peptides.
Plus Tanishq Abraham on Sophont's healthcare world model and Andrew Brackin on Vera Health.
@brackin
@iScienceLuvr
nitter.gq/i/broadcasts/2YREpJMZyâŚ
AI DeSci Guy retweeted
Ginkgo's first autonomous lab sale to a top-tier pharma (Novo) is the real signal, more than any single Q3 line.
Pharma buying the robot fleet for its own site means the bottleneck moved from 'can AI design it' to 'who runs the verification loop in-house'.
@Ginkgo how much of the installed lab's run volume is agent-ordered vs scientist-ordered in year one? Would you like to try more vendors for multiple tasksďź nitter.gq/Ginkgo/status/21187234âŚ
Yesterday, we reported our Q3 2026 earnings. @jrkelly spoke about our first autonomous lab sale to a top-tier pharma company, Novo; how quickly our autonomous labs can grow once theyâre installed; and why frontier AI labs are turning to Cloud Lab for biological data.
Hereâs a clip of him describing how customers can add a new instrument to a Ginkgo autonomous lab in less than a day, much like setting a new piece of equipment on a bench in a manual lab.
Watch the full call here: hubs.la/Q04Dz_Tt0
To see more, read our Q3 results here: hubs.la/Q04DB03b0
AI DeSci Guy retweeted
I still don't understand why everyone is not using this yet. Thanks to it, a year ago I increased my income to 15,000 dollars a month
Andrey Karpathy, co-founder of OpenAI, published a simple idea that got 25 million views: stop using AI to write code, use it to build a second brain
You point Claude Code to a folder, drop any source in there (an article, transcript, PDF) and Claude reads it, links it, and saves it into a living wiki of everything you know. It compounds like interest: the more you feed it, the smarter it gets
Here is the gist:
Install Obsidian, create a vault, open it in Claude Code
Paste the file with Karpathy's wiki idea and tell Claude to build it
Claude creates three folders: raw for sources, wiki for its pages, CLAUDE
which runs everything
Drop any source into raw and say ingest this
Ask questions across everything, forever
5 minutes to set up, and you never start with an empty chat again
The full instructions can be found in the article below
AI DeSci Guy retweeted
Introducing Factory Intelligence.
I've walked through 70+ American factories.
From a distance they look incredible.
Up close, most run on buildings from last century, machines older than your dad, and overworked people holding it all together.
Electrical prefab for data centers is the clearest example: 95%+ of it is still done by hand with power tools, and there aren't enough people to hire.
So we're not bolting robots onto old factories.
We're building new ones from the ground up for autonomy: the cells, the robots, and the software that runs them.
Our first cell is live at an electrical contractor in Indiana.
It learned to bend wire, break tabs and drive screws to build wall outlets, and the pilot became a 3 year production contract expanding to 12 cells.
Next is our own factory: 20 cells with mobile robots moving material between them, building power components for data centers by the end of Q1 2027.
- 220,000 high voltage outlets
-440,000 coils of conduit wire
-36,000 junction boxes
Switch gears next.
AI DeSci Guy retweeted
holy f*ck, 3D DESIGNERS ARE NOT READY FOR WHATâS COMING IN 2027.
your next competitor might not even know how to use Blender.
just ONE person with an AI agent could be building 3D models, materials and entire scenes while youâre still working on the first render.
and the scary part? these tools are already on GitHub.
7 repos every 3D designer NEEDS to know before itâs too late:
1. MCP for Blender - GIVE CLAUDE CONTROL OF BLENDER
let AI create 3D objects, edit materials, build scenes and render.
github.com/ahujasid/mcp-for-âŚ
2. BlendRelay MCP - YOUR AI 3D MODELING ASSISTANT
create and edit Blender scenes using prompts, reference images and AI agents.
github.com/shefq/blendrelay-âŚ
3. TRELLIS.2 - TURN IMAGES INTO 3D
generate detailed 3D assets with complex shapes and realistic PBR materials.
github.com/microsoft/TRELLISâŚ
4. Hunyuan3D - GENERATE 3D MODELS WITH AI
turn reference images into textured 3D objects.
github.com/Tencent-Hunyuan/HâŚ
5. TripoSR - ONE IMAGE TO A 3D OBJECT
reconstruct 3D objects from a single image.
github.com/VAST-AI-Research/âŚ
6. ShaderAgent - LET AI CREATE 3D MATERIALS
an agentic approach to generating procedural materials and shaders.
github.com/VAST-AI-Research/âŚ
7. Nerfstudio - RECONSTRUCT REALITY IN 3D
turn captured images into neural 3D scenes.
github.com/nerfstudio-projecâŚ
the real threat in 2027?
not AI replacing every designer.
itâs someone who knows how to use these tools taking the projects you used to get.
and the BIO-SEALED HEAD youâre looking at?
a surreal human head wrapped in transparent plastic, with an eerie green structure trapped inside.
one of my latest experimental AI visuals.
want to create something like this?
LIKE + BOOKMARK + REPOST.
comment âHEADâ and Iâll send you the prompt.
