Aneesh Sathe ☀️ pfp

Aneesh Sathe ☀️

@ars

257 Following
137 Followers
Aneesh Sathe ☀️ pfp
Aneesh Sathe ☀️
@ars
Manually copy pasting across multiple social networks results in revisions and consequently my least favored networks get the best versions of my thoughts Sorry farcaster, love means you get the least edited 😓
0 reply
0 recast
0 reaction
Aneesh Sathe ☀️ pfp
Aneesh Sathe ☀️
@ars
An LLM tuned to a single individual can be the kernel of new cottage industries
0 reply
0 recast
0 reaction
Aneesh Sathe ☀️ pfp
Aneesh Sathe ☀️
@ars
Ha healthcare is the same but with humans. Much better documentation though :P What would your ideal general purpose KG+ AI tool definitely do?
1 reply
0 recast
1 reaction
Aneesh Sathe ☀️ pfp
Aneesh Sathe ☀️
@ars
What kind of tinkering you thinking about?
1 reply
0 recast
0 reaction
Aneesh Sathe ☀️ pfp
Aneesh Sathe ☀️
@ars
New essay looking at the composite intelligence building material that is expertise+AIAgents+knowledge graphs h/t to @cedric and @vgr for a ton of inspiration https://aneeshsathe.com/2024/05/10/dancing-on-the-shoulders-of-giants/
1 reply
0 recast
2 reactions
Aneesh Sathe ☀️ pfp
Aneesh Sathe ☀️
@ars
A dancer On thoughts disparate Curiosity
0 reply
0 recast
0 reaction
Aneesh Sathe ☀️ pfp
Aneesh Sathe ☀️
@ars
Apparently I draw diagrams to go with blog posts
0 reply
0 recast
0 reaction
Aneesh Sathe ☀️ pfp
Aneesh Sathe ☀️
@ars
This is fine .meme
0 reply
0 recast
2 reactions
Aneesh Sathe ☀️ pfp
Aneesh Sathe ☀️
@ars
We all do babe we all do
0 reply
0 recast
0 reaction
Aneesh Sathe ☀️ pfp
Aneesh Sathe ☀️
@ars
TIL Wright’s Law is a thing. It says that the more of something is made the more efficient individuals and organizations get at it. That rate of getting better is called the learning rate. Now I’m wondering how AI agents can increase how prolific a person can be and at the same time keep the learning rate high.
0 reply
0 recast
0 reaction
Aneesh Sathe ☀️ pfp
Aneesh Sathe ☀️
@ars
I see my nose for tech was again on the right scent, new paper and demo from MS on knowledge graphs w/ RAG. Not using inherent structure in technical & medical fields is silly. now we can leverage literally thousands of years of knowledge. https://arxiv.org/abs/2404.16130 https://www.youtube.com/watch?v=r09tJfON6kE
0 reply
0 recast
0 reaction
Aneesh Sathe ☀️ pfp
Aneesh Sathe ☀️
@ars
Yeah KGs feel like RAG’s older sibling… bit of a pain to setup… I’ll explore if agents can help with that. Have seen a few people post about that
0 reply
0 recast
1 reaction
Aneesh Sathe ☀️ pfp
Aneesh Sathe ☀️
@ars
been exploring agents and AI, was a little disappointed. Tried out knowledge graphs along genAI specifically in the healthcare context.. that seems interesting. Wrote a little companion blog post: https://aneeshsathe.com/2024/05/03/reimagining-ai-in-healthcare-beyond-basic-rag-with-fhir-knowledge-graphs-and-ai-agents/
1 reply
0 recast
1 reaction
Aneesh Sathe ☀️ pfp
Aneesh Sathe ☀️
@ars
Oh thanks for the mention 🙏 Planning to do something much more involved soon :D
0 reply
0 recast
1 reaction
Aneesh Sathe ☀️ pfp
Aneesh Sathe ☀️
@ars
Do protocols dream of Directed Acyclic Graphs?
0 reply
0 recast
3 reactions
Aneesh Sathe ☀️ pfp
Aneesh Sathe ☀️
@ars
0 reply
0 recast
0 reaction
Aneesh Sathe ☀️ pfp
Aneesh Sathe ☀️
@ars
0 reply
0 recast
0 reaction
Aneesh Sathe ☀️ pfp
Aneesh Sathe ☀️
@ars
Delving (hehe) into AI agents more today. These two wrote code for logistic regression (on a toy problem) and then wrote a blog post about it. https://ai.aneeshsathe.com/Two%20Agent%20Task%20Decomposition/Task%20Decomposition%20Using%20Two%20Agents.html
0 reply
0 recast
0 reaction
Aneesh Sathe ☀️ pfp
Aneesh Sathe ☀️
@ars
Does everyone have a screenshots folder??
0 reply
0 recast
0 reaction
Aneesh Sathe ☀️ pfp
Aneesh Sathe ☀️
@ars
Does everyone have a screenshots directory??
0 reply
0 recast
0 reaction