Mark Pors 🦖

5.8K posts

Mark Pors 🦖 banner
Mark Pors 🦖

Mark Pors 🦖

@pors

Building https://t.co/pRJXNjbTGC. High-signal academic paper feeds for agent-assisted research and humans 🦖.

The Hague, Netherlands Bergabung Ekim 2007
1.7K Mengikuti1K Pengikut
Mark Pors 🦖 me-retweet
signüll
signüll@signulll·
one of the most interesting things about ai products today is that almost none of them are *live*. there’s nothing running continuously, reacting to context as it changes.. maybe a scheduled digest here or a timer there, but that’s just pull dressed up as push. everything is fundamentally a vending machine where you walk up, ask, get an answer, & then leave. getting this right is obviously tricky & the business model behind must fit to justify the burn but this is where really interesting application layer problems live rn.
English
152
50
930
125.6K
Mark Pors 🦖 me-retweet
Mark Pors 🦖
Mark Pors 🦖@pors·
Yeah, that is something you don't see LLMs say a lot. They always have something to answer. I found this paper describing a system that encourages an LLM to say "I don't know" when it doesn't have enough confidence to answer the question. Read on 👇
Mark Pors 🦖 tweet media
English
1
2
2
88
Lucas Beyer (bl16)
Lucas Beyer (bl16)@giffmana·
@OfficialLoganK Sorry that i keep repeating this, and i know it's not on you, but you are kinda the face of it so: this ain't Flash progress. Or you need a new brand to mean cheap, because:
Lucas Beyer (bl16) tweet media
English
10
1
190
14.9K
Logan Kilpatrick
Logan Kilpatrick@OfficialLoganK·
Gemini 3.5 Flash has made huge progress from 3.1 Pro on GDPval, Flash is competing at the frontier, post training going strong :)
Logan Kilpatrick tweet media
English
127
39
962
98.8K
Mario Zechner
Mario Zechner@badlogicgames·
when memes go to far: WSJ now says armin co-created pi. fml.
Mario Zechner tweet media
English
30
2
453
68.2K
maithun
maithun@Being_Humor·
Zero calorie Sugar substitutes will bring a pandemic bigger than diabetes.
English
29
49
893
54.7K
Mark Pors 🦖
Mark Pors 🦖@pors·
@jxnlco Please bring EU support up to date wrt computer use and the chrome plugin
English
0
0
3
62
jason
jason@jxnlco·
If you're using codex desktop app today, what features do you feel like are still missing? Let me know and I’ll summarize all the feedback and share internally.
English
934
13
574
72.7K
Mark Pors 🦖
Mark Pors 🦖@pors·
I'm so fed up with the EU deciding for us what we can do and not in terms of privacy. Stop treating us like children. When you make a law/regulation, think first: can the average person make an informed decision here? 1. Companies trying to put bad stuff in our food => as citizens we have no idea what is in there, so yes please: regulate! Well done EU! 2. Deciding if we can use the Codex Chrome plugin => I understand what that can do, and so do most people. OpenAI also warns plenty when using it. Don't regulate this. This puts us further behind tech companies in the RoW.
English
0
0
0
28
Mario Zechner
Mario Zechner@badlogicgames·
someone at the foundation doesn't seem to like me. first they stopped sponsorship (which was distributed to pi contributors, downstream projects, and covered token costs for e2e tests), now i'm no longer part of the GH org. end of an era i guess :) fine tho, i haven't really contributed anything meaningful directly to the repo. and the less people have access the lower the chances openclaw users get shai huluded.
Mario Zechner tweet media
English
52
15
873
99.7K
Mark Pors 🦖
Mark Pors 🦖@pors·
@joshwoodward @ai_for_success You might want to look into this policing behavior in general. It sometimes feels like we are back in school, with the teacher telling us what is good and what is not. I understand the need for guardrails, but you can overdo it.
English
0
0
0
467
Josh Woodward
Josh Woodward@joshwoodward·
@ai_for_success Weird, that shouldn't be happening. DM'd you to see if we can get the underlying videos, and we'll investigate here: b/515000564
English
19
0
142
9.4K
AshutoshShrivastava
AshutoshShrivastava@ai_for_success·
Gemini Omni Flash prompt rejection is crazy 😐 Am I the only one facing this?
AshutoshShrivastava tweet mediaAshutoshShrivastava tweet media
English
83
9
385
75.4K
Mark Pors 🦖
Mark Pors 🦖@pors·
@scaling01 If I read your charts correctly, Gemini 3.1 pro seems smarter and cheaper than GPT-5.5-medium
English
0
0
3
299
Lisan al Gaib
Lisan al Gaib@scaling01·
GPT-5.5-medium has lower end-to-end latency, uses less tokens and is overall smarter and cheaper than Gemini 3.5 Flash it might genuinely be over for anyone not named OpenAI or Anthropic
Lisan al Gaib tweet media
Lisan al Gaib@scaling01

some more Gemini 3.5 Flash benchmarks by Artificial Analysis AI: - the APEX-Agents-AA score is excellent - expected higher on CritPt - reasoning efficiency could also be better, but it kind of depends on what setting they used. if it's the max then it's very good - Price/Perf looks pretty bad vs GPT-5.5. You can get GPT-5.5-medium with better performance for less and with faster responses

English
84
75
1.5K
180.2K
Mark Pors 🦖
Mark Pors 🦖@pors·
@garrytan Are you still considering schema packs as well? I added an issue a while ago.
English
0
0
0
49
Garry Tan
Garry Tan@garrytan·
Dynamic skills are one of the coolest and most powerful parts of the new way to make personal AI work Just in time and markdown is code, and the agent can just change it when you discover new cases to handle Just in time personal software is the most powerful idea of 2026
Marcus@MarcusSpillane

The skillpack architecture is the right call. We run something similar where each skill bundle carries its own tests and the agent can modify them in-flight. The part people miss: letting the agent update its own tooling is what creates the compounding effect. Static skill libraries plateau fast.

English
45
30
390
50.4K
Alon Michael
Alon Michael@SpiceP0dcast·
I just gifted the book to a German lady (wife of my mentor and angel investor) who thinks Elon is the devil. I asked her to read it with an open mind as possible. Not European myself (although I live across the EU for 9 years), the wrong perception of Elon they have here is crazy to me.
English
3
0
16
2.5K
Eric Jorgenson 📚 ☀️
Eric Jorgenson 📚 ☀️@EricJorgenson·
I can tell you which countries are cooked based on how they respond to The Book of Elon. France, Italy, other EUs: "Everyone hates Elon, this book will never sell. Pass." Taiwan, China, India: "This is the largest offer I've ever seen in my 20-year career."
English
53
27
856
64.3K
High Performance Founder
High Performance Founder@hpfounder·
The goal body fat percentage for every guy should be 12-15% body fat. This gives you a foundation for healthy body composition while allowing a decent amount of diet flexibility.
English
12
5
148
20K
Matt Van Horn
Matt Van Horn@mvanhorn·
Use cli’s and a Macmini for OpenClaw or Hermes with chrome for Mac on your main computer? @ me your GitHub handle might have something secret for you to test. Bonus points if you use @ppressdev clis
English
34
2
41
7.9K
Mark Pors 🦖
Mark Pors 🦖@pors·
@deployedmind Yeah, I enjoyed the first part of the book as well. From there on, it was a story about AI safety mostly.
English
0
0
0
12
DeepMind Watch
DeepMind Watch@deployedmind·
I can't help but feel that the book barely dug beneath the surface. I was looking forward to this book but forced myself to the finish line on this one. Imo it's a collection of well-known and already documented stories and just enough new, unpublished details to justify its length. The first part which is tracking Demis's early life was interesting, but the narrative eventually gets bogged down in dry corporate bureaucracy and a revolving door of minor characters. What annoyed me was the book's lack of a critical lens. It felt a bit too 'safe' which makes sense, considering Mallaby admitted in a promo interview that Demis had him speak with his lawyers to restrict certain details from being published. For a biography about a Nobel laureate, Elon Musk’s name is dropped over 150 times and you learn that Larry Page had to be reminded that hummingbirds and whales are cool.
English
1
0
0
33
Mark Pors 🦖
Mark Pors 🦖@pors·
Did anyone else not make it to the end of the @demishassabis biography? Pretty impressive to write a book about someone so fascinating so boring.
English
1
0
1
63
aitization 𝕏 
aitization 𝕏 @aitization·
how do you keep up with the latest AI / ML research papers? you don’t 😎
English
1
0
1
31
Mark Pors 🦖
Mark Pors 🦖@pors·
@tunguz Cool, this will be the first research report on the site I'm going to make: gitxiv.org or grxiv.org. Now I have to figure out the discovery of more reports/papers to solve the discovery problem for others :)
English
0
0
0
5

Jelajahi

@giffmana @OfficialLoganK @badlogicgames @LifeMathMoney @Being_Humor @jxnlco @joshwoodward @ai_for_success