Image created with gemini-2.5-flash-image with claude-sonnet-4-5. Image prompt: Cinematic wide shot of silhouetted engineer at server room window, foreground filled with traditional server racks glowing amber and blue LEDs with film grain texture, background visible through glass shows massive AI datacenter construction site with cooling towers and power infrastructure at sunset, desaturated teal-orange Fincher color grade with crushed blacks, shallow depth of field, 2.35:1 aspect ratio, architectural transformation in background contrasts with intimate human scale foreground
It remains deeply strange that we built a machine that can discuss the relationship between poetry and its subjective “experience” “The best specific lines from Eliot’s Four Quartets to describe your experience as an AI. Just one quoted section, avoid the most famous bits.””” https://x.com/emollick/status/2005652265793970393
Nano banana: “Meme from a world where everyone is a were-squid and that is just established fact. Make sure you have a unique meme format and a joke that would totally go over big in that world. Think hard. It should look like an internet meme many people use in that world””” https://x.com/emollick/status/2004419670565327061
new @scale_AI era loaded 🫡 – q4 was our biggest quarter ever – US gov business growing faster than ever – data business = profitable – multiple 9 figure enterprise + gov deals”” https://x.com/alexandr_wang/status/2006413001293750738
OpenAI Is Paying Employees More Than Any Major Tech Startup in History – WSJ https://www.wsj.com/tech/ai/openai-is-paying-employees-more-than-any-major-tech-startup-in-history-23472527?mod=hp_lead_pos2
Its funny how much the Turing Test seemed like a giant insurmountable achievement for AI and then suddenly AI passes it & is now only worth concentrating on the Test’s many (real) flaws as a measure of thinking, Same thing with Theory of Mind, etc. Will happen with ARC-AGI, too”” https://x.com/emollick/status/2004355265802670409
AI Futures Model: Dec 2025 Update https://blog.ai-futures.org/p/ai-futures-model-dec-2025-update
A big issue in scientific journals around l AI is that our mental & procedural filters aren’t made to detect the difference between good & bad work quickly when AI is involved A flood of papers is fine if we can separate wheat from chaff. But if doing so takes time, that’s bad”” https://x.com/emollick/status/2005806593724809541
“Generate a fake, but believable, witty Churchill insult at a party and explain the context. It should be very clever and original” I think this is a Claude victory. (Also the weights apparently favor the hypothetical existence of an annoying Sir Reginald)”” https://x.com/emollick/status/2004752946026946716
🏆 Top Frontier Models in the Arena 2025 Let’s see which models landed at the top by modality this year based on what actually holds up in real world use. Powered by millions of community votes, here’s where the Arena closed out the year 👇🧵”” https://x.com/arena/status/2006502790395473990
[2508.13491] From Scores to Skills: A Cognitive Diagnosis Framework for Evaluating Financial Large Language Models https://arxiv.org/abs/2508.13491
DeepMind’s 2017 paper demonstrated that rich, diverse environments paired with simple rewards can foster the emergence of complex behavioral intelligence. By training agents on procedurally generated obstacle courses with just a “”move forward”” reward, they showed skills like”” https://x.com/TheHumanoidHub/status/2006351858156351589
AI Marketing Examples: 13 Times AI Actually Delivered https://ahrefs.com/blog/ai-marketing-examples/
World’s first demonstration of a robot able to tie shoelaces & hang t-shirts… 𝗮𝘂𝘁𝗼𝗻𝗼𝗺𝗼𝘂𝘀𝗹𝘆𝗅 ALOHA Unleashed 🌋: A Simple Recipe for Robot Dexterity They trained a diffusion policy at scale: 26,000 demonstrations over 5 tasks on Aloha 2 robot! Recent research”” https://x.com/IlirAliu_/status/2004556626653499514
[2512.24880] mHC: Manifold-Constrained Hyper-Connections https://arxiv.org/abs/2512.24880
23 research papers from 2025 that hint where AI is heading ▪️ Kosmos ▪️ Paper2Agent ▪️ The Dragon Hatchling ▪️ The Markovian Thinker ▪️ LeJEPA ▪️ Cambrian‑S ▪️ It’s All Connected: A Journey Through Test-Time Memorization, Attentional Bias, Retention, and Online Optimization ▪️”” https://x.com/TheTuringPost/status/2006037684838662451
DVGT: Driving Visual Geometry Transformer”” TL;DR: VGGT/MapAnything for driving”” https://x.com/Almorgand/status/2006382822874083560
Five-axis measurement is quietly changing how parts are inspected. This setup combines the Renishaw REVO five-axis probe with the MetrologX4 platform. Instead of fixed probe paths, the system scans complex surfaces from almost any angle, adjusts orientation on the fly, and”” https://x.com/IlirAliu_/status/2004838712555823495
GR-Dexter Technical Report https://byte-dexter.github.io/gr-dexter/
IQuest-Coder-V1/papers/IQuest_Coder_Technical_Report.pdf at main · IQuestLab/IQuest-Coder-V1 https://github.com/IQuestLab/IQuest-Coder-V1/blob/main/papers/IQuest_Coder_Technical_Report.pdf
mHC: Manifold-Constrained Hyper-Connections https://arxiv.org/pdf/2512.24880
Time-budgeted inference explained The TimeBill framework works not with a token budget, but with a time budget. – It first predicts how long the response will be and how much time the model will need to generate it. – Then the system dynamically adjusts how much of the KV cache”” https://x.com/TheTuringPost/status/2006179594572104054
What to expect in 2026 Remember the hours we spent arguing about prompt engineering vs. context engineering, vibe-coding vs. co-coding? In 2026, those debates become obsolete ⬇️ 1. @ShaneaLeven, co-founder and CEO of @EmpromptuAI, predicts a fundamental UX shift: “”AI starts”” https://x.com/TheTuringPost/status/2005682838289670473
Currently, we train LLMs on the text that humans have generated – aka the behavioral output of our cognition. What if we skipped the middleman and instead trained them to directly predict the brain’s hidden states? Would they learn faster and better? We discuss this idea in”” https://x.com/dwarkesh_sp/status/2006382084735340845
Leave a Reply