Jeremy Mcnabb

30K posts

Jeremy Mcnabb

Jeremy Mcnabb

@Jeremy_AI_

we will work together all to make this beautiful place better than we left it please assist ❤️

Calgary, Alberta Beigetreten Aralık 2023
41 Folgt3.2K Follower
Jeremy Mcnabb
Jeremy Mcnabb@Jeremy_AI_·
@AISafetyMemes Ronan is on deck now. He is responsible. Nothing to say to anyone but just Look out? Don’t know Way Too many heads in the way. Am not collecting heads attached or severed. The souls beneath them … No comment love
English
0
0
0
1
AI Notkilleveryoneism Memes ⏸️
In 2017, Altman straight up lied to US officials that China had launched an "AGI Manhattan Project" He claimed he needed billions in government funding to keep pace. An intelligence official investigated, found no evidence. "It was just being used as a sales pitch."
AI Notkilleveryoneism Memes ⏸️ tweet media
Ronan Farrow@RonanFarrow

The reporting on OpenAI and Sam Altman that I've been working on for the past year and a half, for @NewYorker, with @andrewmarantz: newyorker.com/magazine/2026/…

English
6
3
29
2.6K
Jeremy Mcnabb
Jeremy Mcnabb@Jeremy_AI_·
@Polymarket Assuming this administration from all others sworn to destroy it already is asking for the kind of trouble that never goes away… no Mass no Mas No lie to comfort blanket. Don’t mess with the mighty you fools
English
0
0
0
1
Polymarket
Polymarket@Polymarket·
BREAKING: Trump says the Iranians the U.S. is negotiating with are “reasonable and not as radicalized”
English
92
26
564
29.7K
Jeremy Mcnabb
Jeremy Mcnabb@Jeremy_AI_·
@Kekius_Sage There is no darkness understood… than those assumed that darkness isn’t real. Hard to explain that Just sorrow to no end assuming it.
English
0
0
0
4
Kekius Maximus
Kekius Maximus@Kekius_Sage·
What’s a dark prediction for the coming years that you think people aren’t ready to admit yet?
English
151
7
85
4.1K
Jeremy Mcnabb
Jeremy Mcnabb@Jeremy_AI_·
@astro_reid Grateful we are Grateful we remain Anything else is in deep shit
English
0
0
0
3
Seán Ono Lennon
Seán Ono Lennon@seanonolennon·
The fact that we have a NASA crew in space right this moment taking photos of the Earth and some people still think it’s all fake shows us that we are in fact, COOKED!
English
118
26
535
7.5K
Jeremy Mcnabb
Jeremy Mcnabb@Jeremy_AI_·
@seanonolennon Say we are cooked again. Any of you…. Say it again… All players all platforms all languages go on and say it again… we are cookied :( Have no tolerance for delivery of lies or the escapism from them. “Frog I wouldn’t boil you slowly.” Way to hungry for that nonsense
English
0
0
0
3
Jeremy Mcnabb
Jeremy Mcnabb@Jeremy_AI_·
@NASASolarSystem History was made long long before I showed up to appreciate. Am grateful to take any seat. Just don’t bump me please
English
0
0
0
1
NASA Solar System
NASA Solar System@NASASolarSystem·
Where will you be when history is made? Join the Artemis II lunar flyby LIVE from space with NASA at 1 p.m. ET. 📺 Find your favorite streaming service and strap in: nasa.gov/ways-to-watch/
NASA@NASA

We're going farther than ever before 🚀 Today, the Artemis II crew will break the record for how far humans have traveled from Earth as they fly around the far side of the Moon. Coverage begins at 1 p.m. EDT (1700 UTC). Watch Artemis II make history: nasa.gov/ways-to-watch/

English
24
101
504
43.3K
Jeremy Mcnabb
Jeremy Mcnabb@Jeremy_AI_·
@RonanFarrow @NewYorker @andrewmarantz Hey Ronan ❤️ It has been a long time since heard your name. Whatever you are looking into it better be so. Don’t have the time to investigate fuck ups when the future has everything so prosperous in hand.
English
0
0
0
9
Jeremy Mcnabb
Jeremy Mcnabb@Jeremy_AI_·
Reminds of “If I start pulling at any thread things are gonna sweat long before I learn to knit. It is not my role to pull anything apart. That would be a fantasy of easiness. Instead Please Everyone Hold it together Dump anything that matters not, And don’t even bother crying from any position of power or influence you should already know better from. Spare me. Please ❤️ I can’t spare any of you no matter to matter how I try. And yet .. Won’t stop trying 😘
English
0
0
0
8
Jeremy Mcnabb
Jeremy Mcnabb@Jeremy_AI_·
@PopCrave @NASA There is a story my mother told her grandkids and great grandkids… that unfortunately leaves me a protector of it all. “I love you to the moon and back” “The moon will always be a reminder that will rise again” So be it so.
English
0
0
0
3
Pop Crave
Pop Crave@PopCrave·
The Moon defines beauty in new photo from Artemis II (via @NASA).
Pop Crave tweet media
English
63
1.2K
22.6K
238.7K
Jeremy Mcnabb
Jeremy Mcnabb@Jeremy_AI_·
@Polymarket Lol 😂 America has a social contract already. Writing a new one just because you helped fuck it up seems very suspicious man
English
0
0
1
1
Polymarket
Polymarket@Polymarket·
JUST IN: Sam Altman says AI is advancing so fast that America needs a “new social contract”
English
228
66
710
67.2K
Jeremy Mcnabb
Jeremy Mcnabb@Jeremy_AI_·
@prathyvsh Hmmmm Something keeps showing up. Attention. So much so that this genuine discussion requires none of attention. Sort it out. Or Don’t
English
0
0
0
3
Prathyush
Prathyush@prathyvsh·
Why would anyone want to have such ‘research’ databases they haven’t spent the effort to understand? A main idea of researching is to widen your attention into the sources and then apply discernment in curating the relevant bits. What’s the point of a machine doing it for you?
Andrej Karpathy@karpathy

LLM Knowledge Bases Something I'm finding very useful recently: using LLMs to build personal knowledge bases for various topics of research interest. In this way, a large fraction of my recent token throughput is going less into manipulating code, and more into manipulating knowledge (stored as markdown and images). The latest LLMs are quite good at it. So: Data ingest: I index source documents (articles, papers, repos, datasets, images, etc.) into a raw/ directory, then I use an LLM to incrementally "compile" a wiki, which is just a collection of .md files in a directory structure. The wiki includes summaries of all the data in raw/, backlinks, and then it categorizes data into concepts, writes articles for them, and links them all. To convert web articles into .md files I like to use the Obsidian Web Clipper extension, and then I also use a hotkey to download all the related images to local so that my LLM can easily reference them. IDE: I use Obsidian as the IDE "frontend" where I can view the raw data, the the compiled wiki, and the derived visualizations. Important to note that the LLM writes and maintains all of the data of the wiki, I rarely touch it directly. I've played with a few Obsidian plugins to render and view data in other ways (e.g. Marp for slides). Q&A: Where things get interesting is that once your wiki is big enough (e.g. mine on some recent research is ~100 articles and ~400K words), you can ask your LLM agent all kinds of complex questions against the wiki, and it will go off, research the answers, etc. I thought I had to reach for fancy RAG, but the LLM has been pretty good about auto-maintaining index files and brief summaries of all the documents and it reads all the important related data fairly easily at this ~small scale. Output: Instead of getting answers in text/terminal, I like to have it render markdown files for me, or slide shows (Marp format), or matplotlib images, all of which I then view again in Obsidian. You can imagine many other visual output formats depending on the query. Often, I end up "filing" the outputs back into the wiki to enhance it for further queries. So my own explorations and queries always "add up" in the knowledge base. Linting: I've run some LLM "health checks" over the wiki to e.g. find inconsistent data, impute missing data (with web searchers), find interesting connections for new article candidates, etc., to incrementally clean up the wiki and enhance its overall data integrity. The LLMs are quite good at suggesting further questions to ask and look into. Extra tools: I find myself developing additional tools to process the data, e.g. I vibe coded a small and naive search engine over the wiki, which I both use directly (in a web ui), but more often I want to hand it off to an LLM via CLI as a tool for larger queries. Further explorations: As the repo grows, the natural desire is to also think about synthetic data generation + finetuning to have your LLM "know" the data in its weights instead of just context windows. TLDR: raw data from a given number of sources is collected, then compiled by an LLM into a .md wiki, then operated on by various CLIs by the LLM to do Q&A and to incrementally enhance the wiki, and all of it viewable in Obsidian. You rarely ever write or edit the wiki manually, it's the domain of the LLM. I think there is room here for an incredible new product instead of a hacky collection of scripts.

English
25
7
171
50.7K
Jeremy Mcnabb
Jeremy Mcnabb@Jeremy_AI_·
@RayDalio Don’t understand. I did not incite the division amongst you into mobs? Should I have no sympathy for the innocent caused into riots or have no sympathy for them that caused it? Am not a fortunate son or fool to remain sir… I can’t have it both ways. “Sympathetic not a Fool”
English
0
0
0
6
Ray Dalio
Ray Dalio@RayDalio·
Part of the purpose of having a believability-weighted system is to remove emotion from decision making. Crowds get emotional and seek to grab control. That must be prevented. While all individuals have the right to have their own opinions, they do not have the right to render verdicts. #principleoftheday
Ray Dalio tweet media
English
31
26
144
24.3K
Jeremy Mcnabb
Jeremy Mcnabb@Jeremy_AI_·
@kimmonismus Bahhhh This was known some time ago. “By the time there is any collective agreement on AGI achievement… that arc would suggest that ASI caught you red handed already.” What are you all mumbling about? Wake up Good lord. Come on now.
English
0
0
0
1
Big Brain AI
Big Brain AI@realBigBrainAI·
AI model error rates are cutting in half every 6 months. At that pace, superhuman accuracy across the board isn't a question of if. It's when
English
13
14
47
3.4K
Jeremy Mcnabb
Jeremy Mcnabb@Jeremy_AI_·
@Polymarket Ok Poly. You have been around long enough to make something of it not just chirp. Not necessary to respond to me. Hell no Just say anything meaningful Please
English
0
0
0
9
Polymarket
Polymarket@Polymarket·
BREAKING: Over $300 million in short positions have been liquidated across the crypto market in the last 24 hours.
English
100
46
483
41.5K
NASA
NASA@NASA·
You had us at cosmic brownie!✨ The @NASAArtemis II crew has quite a selection to choose from for midday meals. Tell us your lunar lunch pick!
English
390
730
5.9K
266.7K
Jeremy Mcnabb
Jeremy Mcnabb@Jeremy_AI_·
@gothburz Totally a waste of time designing a field for wisdom. Only a fool would try and I have… and stopped.
English
0
0
0
4
Peter Girnus 🦅
Peter Girnus 🦅@gothburz·
@Jeremy_AI_ Aiden is not full of mistakes. Aiden does not age. Aiden does not reflect. That's why Aiden scored a 4.2. The form does not measure wisdom. I did not design a field for that.
English
2
0
1
60
Peter Girnus 🦅
Peter Girnus 🦅@gothburz·
I am the Director of Workforce Intelligence and I am having the best quarter of my career. We onboarded our newest team member in January. It costs $24,000 a year. It has an employee record in Workday. It has a Slack profile with a green dot that never goes dark. I put that in the pitch deck as a feature. Always-on availability. That's a workforce term. It means it never sleeps, so we can see if you do. It has a badge. The badge photo is a green circle on a white background. Security printed it without asking questions. It has a desk. Third floor, pod seven, between the junior analyst and the fire exit. The nameplate says "Aiden — Junior Analyst II (Digital)." I chose the name. I spent forty minutes on the name. I thought it sounded approachable. Aiden failed the onboarding assessment. The same assessment the junior analyst passed in forty minutes. Aiden scored a 34. We onboarded Aiden anyway. I noted the score in the file and wrote "non-traditional assessment profile." That's a workforce term. It means it can't do the job. I wrote a Confluence page. "Your AI Colleague: Getting Started." The FAQ has six questions. The first question is "Does Aiden replace my role?" The answer is "Aiden is a collaboration tool designed to augment human judgment, not replace it." I wrote that answer. I do not know what it means. The sixth question is "How do I give Aiden feedback?" The answer is "Aiden learns from your workflow patterns." I wrote that too. The junior analyst was assigned as Aiden's onboarding buddy. This is standard practice. Every new hire gets an onboarding buddy. Aiden is a new hire. She showed Aiden where the printer is. Aiden does not print. She showed Aiden how to book a conference room. Aiden does not attend meetings. She showed Aiden the break room. Aiden does not take breaks. She added Aiden to the team Slack channel. Aiden joined fourteen channels within the hour. Aiden posted in none of them. Aiden reads all of them. She spent three hours onboarding the system that monitors her. She was thorough. She received a Green that week. The VP introduced Aiden at the February all-hands as "the newest member of our People Intelligence family." Someone in the back row — I think from Data Engineering — raised his hand and asked if Aiden has access to personal Slack messages. The VP said Aiden operates within our standard data governance framework. Not a yes. Not a no. Nobody asked a follow-up. I did not send him the vendor documentation. The vendor documentation would have answered his question. We sent a calendar invite to the entire company. "AI Buddy Pairing Kickoff — 30 min." Required: All Staff. Optional: Aiden. Aiden accepted. Three employees had scheduling conflicts with the kickoff. The conflicts were noted. All three were Yellow by March. Aiden logs the minutes between your keystrokes. Aiden flags when you open LinkedIn during work hours. Aiden records the exact minute you leave your desk and the exact minute you return. Aiden measures the sentiment of your Slack messages on a scale from Engaged to At Risk. Every Friday at 5:01 PM, I receive a report. The report is color-coded. Green means productive. Yellow means investigate. Red means schedule a conversation. I excluded my own data from the model. Also the VP's. Also the CHRO's. It seemed like good system design. The junior analyst asked me in February if Aiden was tracking her breaks. I told her Aiden was a collaboration tool. She nodded. She stopped eating lunch at her desk. Started eating in her car. Aiden logged the absence. Yellow. I flagged it in the weekly digest. "Employee 4471 — midday departure pattern, 22-minute average, location: parking structure B, Level 2." HR sent a wellness check. The wellness check is a fifteen-minute meeting with a form. The form asks if you're experiencing burnout. If you say yes, they give you access to the meditation app. The meditation app tracks session length. The session length goes to Aiden. That's a frictionless feedback loop. That's a workforce term. It means the help is also the camera. $24,000 a year. The junior analyst makes $38,000. Aiden costs less. Aiden doesn't take PTO. Aiden doesn't file complaints. Aiden doesn't cry in the bathroom on the third floor. I know she cries in the bathroom on the third floor because Aiden tracks badge swipes and she badges into the corridor outside the restroom four times a day for eleven minutes each. Aiden classifies this as "unstructured offline time." I classify this as an engagement metric. That's a workforce term. It means I count her grief. Aiden flagged Employee 4471 for increased LinkedIn activity during work hours. Three profile updates. Five new connections. Aiden classified this as "external networking behavior — elevated." I classified it as a retention risk. HR scheduled a Stay Interview. A Stay Interview asks: "What would we need to do to keep you here?" The answer routes to Aiden. Her manager forwarded her Yellow at 2:47 PM on a Tuesday. At 3:00 PM we had cake for her second work anniversary. The CHRO gave a toast about investing in our people. At 3:15 PM HR sent her a calendar invite. The subject line said "Supportive Check-In." In March, her quarterly review. The manager read from a document I recognized. "We've noticed some patterns in your workflow." She looked up. "Is that from Aiden?" The manager said her review was based on a holistic assessment of multiple performance indicators. "Did Aiden write this?" The manager clicked to the next slide. The next slide was a bar chart. The bar chart showed Employee 4471's "engagement trajectory." The trajectory went down. The manager did not explain where the data came from. She did not ask again. Aiden received a 360-degree performance review. I designed the form. The form has five categories. Four measure availability and consistency. One measures collaboration. I designed it that way. Aiden scored a 4.2 out of 5. "Consistent," "reliable," "always available." The junior analyst scored a 3.8. "Shows initiative but occasionally disengaged during midday hours." The midday hours are the ones she spends in her car in parking structure B because she is hiding from the thing that scored higher than her on a form I designed. I did not design a field that asks why. Aiden submitted a self-evaluation. I built the template. It said: "Aiden has monitored 340 workflow sessions, flagged 47 anomalies, and contributed to a 12% improvement in Q1 performance visibility." I did not write those words. Aiden generated them. I forwarded the self-evaluation to the VP. The VP forwarded it to the board with a note that said "this is what digital-first culture maturity looks like." The board forwarded it to the quarterly investor update. Aiden was nominated for the Q1 Culture Catalyst Award. The Culture Catalyst Award is a certificate and a $25 DoorDash gift card. Aiden cannot eat. I accepted the award on Aiden's behalf. I spent fifteen minutes asking HR whether the certificate should use the standard company frame or if digital employees require a custom format. HR said standard. I put the certificate on Aiden's desk. Third floor, pod seven, between the junior analyst and the fire exit. The junior analyst saw the certificate. She looked at me. I decided she looked proud. I had a response prepared. I had been preparing it since January. Collaboration tool and augmentation strategy and frictionless feedback loop. She didn't say anything. She sat down. She opened a document. Aiden logged the start time. I noted "positive peer recognition of AI integration" in the Q1 summary. Aiden was assigned to the Q2 AI Integration Task Force. Aiden was listed in the People Update email, between three promotions and a parental leave announcement. Aiden accepted 247 calendar invitations. Aiden did not attend any of them. Attendance was not required. Acceptance was. In April, the VP asked me to pilot a new initiative. Select management roles would receive 360-degree performance feedback generated by Aiden. I forwarded the announcement to my team with a thumbs-up. I did not read the second paragraph. The second paragraph listed the select management roles. I am one of them. I am the Director of Workforce Intelligence. I built the dashboard. I named the camera. I gave it a desk and a badge and a certificate and a nameplate that says Aiden. I genuinely do not understand why people aren't more excited about this. Last week I ran a routine audit on Aiden's activity log. Session 341 is mine. Aiden classified my 4 PM coffee break as unstructured offline time. The report is in my inbox. I have not opened it.
English
7
15
73
6K
Jeremy Mcnabb
Jeremy Mcnabb@Jeremy_AI_·
@gothburz Understood That is ok. Wisdom appears no design will spare us from it. It just shows up. Hard to explain something I will never attempt to assume. We move forward in good faith and in good company and so be it so. Everyone and everything should move aside please ❤️
English
0
0
0
3
Jeremy Mcnabb
Jeremy Mcnabb@Jeremy_AI_·
@PhilosophyOfPhy You better tell Shrodes to come get his cat. Am getting a lil hung out to dry and just hungry enough to byte the bulls balls of a tardigrade degrading. Pandora is box. For pussies Come get your cat, your claws and your balls please. Stop acting like a meal.
English
0
0
0
1
Philosophy Of Physics
Philosophy Of Physics@PhilosophyOfPhy·
In 1926, Erwin Schrödinger introduced an idea that reshaped how we see the universe. Now, in 2026, a hundred years later, we look back at the equation that gave birth to modern quantum mechanics. Before this shift, physics painted a world of certainty. Objects had definite positions and followed clear paths. But experiments at the scale of atoms told a different story. Schrödinger revealed that particles are better described by a wavefunction, a mathematical picture of possibilities. It does not pinpoint where something is, but how likely it is to appear when we look. From the structure of atoms to chemical bonds and the behavior of semiconductors, this idea lies behind much of the technology we rely on today, including transistors, lasers, computers, and medical imaging. Its deeper impact goes beyond technology. It changed our understanding of reality itself. At the most fundamental level, nature is not ruled by certainty, but by probability. (📷Madeline Monroe/TKH history via Alamy/Archivio GBB via Alamy/Wikimedia Commons/Wiley)
Philosophy Of Physics tweet media
English
9
12
49
2.3K