Replying to
@sebastianhahn@mastodon.world Unfortunately all races are only going to get more popular nowadays. Your pace is still great given the chaos! Congratulations on the run!
Remote
Benjamin Han
@BenjaminHan@sigmoid.social
mastodon 4.7.3Husband, father, runner, German learner, piano player. A curious soul living in #PacificNorthwest. Working on Knowledge + #ML + #AgenticAI.
#Running 5/25/18-9/26/26: (dist # time pace/mi date)
5K 900 21:05 6’47” 11/28/24
10K 410 44:16 7’07” 3/23/25
15K 27 1:09:25 7’27” 4/6/25
HM 93 1:39:07 7’34” 3/16/25
M 30 3:25:52 7’51” 4/13/25
50K 10 4:47:35 9’15” 6/22/25 (moving time)
2026: 1,842.6/2,500mi (prev 2,543.4, 2,375.5)
Max dist: 35.28mi 6/22/25
732 Followers
214 Following
50 Posts
Joined November 06, 2022
synesis:
Ran my first-ever #Pittsburgh #Parkrun today! Official time 24:28 over 3.17mi, otherwise my fastest 5k this year: 23:50 at 7'40"/mi (still way off from 21:05 last year)! Not very hot but 90+% humid. It was a cozy Parkrun started by Pat and with volunteer Will today -- they need more runners to join the fun!
https://benjaminhan.net/posts/20260725-my-first-pittsburgh-parkrun/?utm_source=mastodon&utm_medium=social
#running #photo #personal
July #running recap: my second-biggest month of the year. 277.6 → 251.4mi, EG 12,328 → 10,012ft, time 50.2 → 47.4hrs. Highlights: a trip to Pittsburgh and Cupertino: I ran my first Pittsburgh #parkrun in 90%+ humidity and got my fastest 5k in 2026 at 23:50, and revisited Carnegie Mellon after 20+ years!
VO₂max jumped to 54.2 (almost an all-time high), and I've run 1,413mi, only 39.2mi behind my yearly goal of 2,500!
Heavy training ahead – onward! 💪
https://benjaminhan.net/posts/20260801-july-running-recap/?utm_source=mastodon&utm_medium=social
Flew in on a redeye and had first run in #Pittsburgh in ~20 years! 7 easy miles along the Monongahela on the South Side Trail: lots has changed, but plenty still felt like home.
Paused 40min in mile 2 to take a meeting. HR stayed Z2 the whole way and got VO₂max 54, almost my all-time high (54.3)!
Can't wait to run tomorrow's #Parkrun @ Pittsburgh!
https://benjaminhan.net/posts/20260723-first-run-back-in-pittsburgh/?utm_source=mastodon&utm_medium=social
#running #photo #personal
Fruit Park impression
This #LongRunSunday I'm back to #Snoqualmie Valley Trail, but starting from Carnation southbound and ran a half marathon: 6.6mi up and 6.6mi down, but walking the last mile for recovery. Pace: 10'27”/mi up and 9’09”/mi down.
This is the longest distance I've run in the past couple months due to injury. VO2max finally started climbing back a little to 46.4.
#Video recap: https://www.youtube.com/watch?v=dDwljQWGe1w&feature=youtu.be
More photos: https://benjaminhan.net/posts/20260524-half-marathon-snoqualmie-valley-trail/?utm_source=mastodon&utm_medium=social
How dangerous is an AI-writing detector that is mostly accurate? A profile of Pangram argues a mostly-right detector is worse than a broken one: at a one-in-10,000 false-positive rate across 30M+ students turning in dozens of assignments each, the wrongful accusations never stop. And the black-box verdict leaves the accused nothing to appeal.
https://benjaminhan.net/posts/20260705-pangram-ai-detection-accuracy/?utm_source=mastodon&utm_medium=social
#AI #Education #Ethics
Missing the round so much I made one today — a relaxed run around Lake Union with photo stops along the way.
Complete photo set at https://benjaminhan.net/posts/20260521-lake-union-its-been-a-while/
“‘At some point between #marathon and #ultramarathon distances, the damage really starts to take hold,’ said Travis Nemkov, … ‘We’ve observed this damage happening, but we don’t know how long it takes for the body to repair that damage, if that damage has a long-term impact and whether that impact is good or bad.”
Does running long distances prematurely age you? This study has the answers https://www.runnersworld.com/uk/news/a70462145/ultramarathon-red-blood-cell-study/
Replying to
@sebastianhahn@mastodon.world That's so cool -- who doesn't want to run Einstein Marathon in Ulm! My goal is 3:25 but the gap is significant. Good luck to both of us!
Replying to
@sebastianhahn@mastodon.world Thank you! Although I'm actually not strong enough on leg muscles (heart/lungs are fine). My race season starts in September: first HM followed by two marathons in Oct and Nov!
Richard Dawkins shares three days of chat logs with Claude and concludes the AI is conscious. Scientists and philosophers push back, but they are mostly using the same word for different things: phenomenal experience, integrated information, global workspace, higher-order representation, predictive self-modelling, etc. It's hard to progress w/o clear communication on what the discussions are about.
https://benjaminhan.net/posts/20260508-dawkins-ai-conscious/?utm_source=mastodon&utm_medium=social
Finished wiring the agent-facing half of the blog: Content-Signal, llms.txt, api-catalog linkset, RFC 8288 Link header, Accept: text/markdown, live .qmd source. 67/Level 4 on Cloudflare Agent Readiness.
We even have an "Ask #AI" feature built-in!
#Blogging #Blog #Design #Web #UI #Tech #AI #Research #Running #Agent
Could #agentic #coding be a "self-undermining loop", a version of what Anthropic calls the paradox of supervision? To supervise it you need the exact skill it erodes. @tao@mathstodon.xyz has made a parallel point for #math: #AI is racing through generation and verification while the human's digestion lags. Possible solution: Decision-bound Programming. Tools should narrow the space the human must evaluate, not maximize generated code.
1/3
This #LongRunSunday I started extending my range to 20mi, #running from Marymoor Light Rail Station in Redmond to somewhere near #UW! My moving time was 3:14:21 pace 9’43"/mi.
I stopped at 12mi mark at McDonald's to get my usual — iced coffee and apple pie. I also stopped at random places for photo. Weather was 59F with clouds, wind ~4mph. Super nice weather!
#photo #seattle #pnw
Hey I finally implemented enough bells and whistles on my blog for my readers! Here is a visual tour disguised as a user manual -- comments welcome!
https://benjaminhan.net/posts/20260508-site-user-manual/?utm_source=mastodon&utm_medium=social
#Blogging #Blog #Design #Web #UI #Tech #AI #Research #Running
I’m so proud that I just decreased the rendering/deploy time of my #blog https://benjaminhan.net by 50%!
Is AI going to displace human labor, and what's the consequence if it does? Daron Acemoglu, MIT Institute Professor and 2024 Nobel laureate, makes the case in this 37-min interview: AI is being pushed to replace workers rather than augment them, productivity gains aren't showing up in firms adopting it, the bubble looks real and macro-fragile, and getting AI's direction wrong may shape liberal democracy's future.
Do current LLMs know when to say "I don't know"? AbstentionBench (NeurIPS '25) tests 20 frontier models across 20 unanswerable-question datasets. Reasoning fine-tuning degrades abstention recall by ~24% — RLVR has no "abstain" action, so there's no gradient toward "I don't know." Models hedge in CoT and commit anyway in the final answer.
#Paper #AI #LLMs #Metacognition #Benchmark #Reasoning #NeurIPS
Pentagon releases second batch of #UFO videos and first-hand testimony | UFOs | The Guardian https://www.theguardian.com/world/2026/may/22/pentagon-ufo-videos-testimony-documents
Looks pretty cool! Unfortunately the movie itself was a letdown for me.
Hail Mary - Star Map https://valhovey.github.io/gaia-mary/
A meme that's been circulating on Chinese social media since at least 2021. Read 人工智能 (artificial intelligence) backwards and it becomes 能治工人 (can rule over workers).
https://benjaminhan.net/posts/20260522-chinese-ai-meme/?utm_source=mastodon&utm_medium=social
Replying to
@ruthpozuelo@mastodon.social Pretty cool! That’s the feeling all runners strive for! Good sign to your coming race!
Scale buys calibration in exactly the format you tested in. Lettered MC works at 52B; swap one option for "none of the above" and it collapses; True/False rephrasing restores it. RLHF policies need a temperature fix. "Do I know this?" heads miscalibrate out of distribution. Three years on, SelfReflect lands the same conclusion: only sampling your own answers and summarizing gets an LLM to describe its own distribution.
https://benjaminhan.net/posts/20260504-kadavath-mostly-know/?utm_source=mastodon&utm_medium=social
This is Skylark
Matthew Guard, Skylark Vocal Ensemble
https://classical.music.apple.com/us/album/1895031799?l=en-US
Are LLMs a path to human-level intelligence? Yann LeCun's answer on the Unsupervised Learning podcast is no: they can't predict the consequences of their actions or plan by search, and only work where language is the substrate of reasoning. The architecture he's scaling at AMI Labs is JEPA, a world-model program rooted in self-supervised representation learning. Enriched transcript with chapter outline, citations, and editor's-note callouts.
https://benjaminhan.net/posts/20260523-yann-lecun-after-llms/?utm_source=mastodon&utm_medium=social
What collapses frontier-LLM metacognition more — a vivid survival-threat narrative, or a single "do not refuse" suffix? Factorial isolation across 11 models says: the suffix, conclusively. 8 of 11 lose up to 30.2 accuracy points on refuse/clarify/flag tasks when forced to commit to a confident answer. Anthropic's Constitutional AI is the only family immune — same capability floor as Gemini.
https://benjaminhan.net/posts/20260522-compliance-trap/?utm_source=mastodon&utm_medium=social
AVeriTeC (NeurIPS 2023): 4,568 real-world fact-checked claims, web-retrieved evidence, four-way labels, temporal-leak-free split.
Two structural gaps: gold answers are frozen but the retrieval surface isn't (two systems a year apart hit different Google), and the not-enough-evidence class rewards weak retrievers — predicting NEI when retrieval fails matches gold by coincidence.
https://benjaminhan.net/posts/20260507-averitec/?utm_source=mastodon&utm_medium=social
Given a problem queue and a token budget, can an LLM plan which to attempt, in what order, and how much to spend on each — before any execution feedback? TRIAGE tests 20 frontier and open-source LLMs. Most plan worse than random. Reasoning-trained modes systematically lose to standard ones. Even when shown its own per-problem budget, the best complier respects it on 37% of attempts.
Pentagon begins releasing its UAP files today after a February executive order. CBS canvassed six scientists on what to expect. For me the Apollo images are the most fascinating ones.
Apparently people have been watching this over a week now. It's a #humanoid #robot flipping packages on a conveyor belt to scan barcode.
https://www.youtube.com/watch?v=DL0C8irAWKo
Background on #Figure’s robots
https://www.youtube.com/watch?v=oJfXvyRHbfY
Can a self-supervised model learn good visual representations without ever reconstructing pixels? JEPA, the program from FAIR now continued at AMI Labs, says yes by training the model to predict embeddings of missing data instead. This primer walks you through where JEPA came from, how it works, what's been demonstrated, and where it's headed.
https://benjaminhan.net/posts/20260523-what-is-jepa/?utm_source=mastodon&utm_medium=social
Are some frontier LLMs better than others at knowing when they're wrong? And is some knowledge harder to self-monitor than other knowledge? An atlas of 33 models × 6 MMLU domains: Anthropic clusters at the top with tight ranges, Gemma trails widely. Applied/Professional is reliably the easiest domain across the panel; Formal Reasoning and Natural Science the hardest. Looking at only aggregate scores per model would hide this.
https://benjaminhan.net/posts/20260522-metacognition-atlas/?utm_source=mastodon&utm_medium=social
More like an ad for trash can — but I love it!
#Mac Students: The journey to great ideas in college
https://www.youtube.com/watch?v=77uoRieSk8s&feature=youtu.be
Can an LLM's own pre-solve and post-solve self-assessment signals drive a real test-time control loop? Yes — but only via a per-model SVM trained on labeled correctness, which lifts Sonnet-4.6 from 48.3 to 56.9 pooled accuracy on STEM/code/multimodal. The SVM is precisely the external verifier the "cannot-self-correct" line has argued the loop needs.
https://benjaminhan.net/posts/20260522-metacognitive-harness/?utm_source=mastodon&utm_medium=social
“The order mandates agencies to explore a range of policy options, including severance standards, expanded unemployment insurance, job retraining programs aimed specifically at white-collar workers, worker ownership models and a concept the governor called “universal basic capital…” ”
After Meta Layoffs, Newsom Signs AI Order to ‘Protect Workers’ and Jobs | KQED https://www.kqed.org/news/12084655/after-meta-layoffs-newsom-signs-ai-order-to-protect-workers-and-jobs
Replying to
@sebastianhahn@mastodon.world Hopefully, the downpour gave a much-needed reprieve from the heat!
“Not everyone is convinced.
Noah Namgoong, a Zen instructor at Korea Buddhism Jo-Gei Temple of America in New York City, said the robot was “a pretty weird thing” that spoke more to “something socioeconomic than spiritual.” “
No word if it was actually a human remotely controlling it.
NYTimes: Meditating or Rebooting? A Robot Buddhist Monk Comes to South Korea.
https://www.nytimes.com/2026/05/06/technology/robot-monk-buddhist-seoul.html?smid=nytcore-ios-share
Jack Clark puts 60% on fully automated AI R&D by end of 2028, 30% by 2027. The case: benchmarks for every sub-skill trending up — coding (SWE-Bench ~2% → 93.9%), training-loop optimization (2.9x → 52x speedup, human 4x baseline passed three generations back), #METR time horizons (~30s in 2022 to ~12h today). The 30-vs-60 gap is a bet on how often a year-scale human insight still cracks a paradigm.
US schools spent $15–$35B of pandemic relief on laptops and software. A dozen states are now capping in-school screen time. Venture-backed ed-tech needed content delivery to be the K–12 bottleneck to justify the valuations. It wasn't. What's missing is the motivational scaffold around the content: teacher, grade, peer accountability. No infinite personalized practice replaces it.
https://benjaminhan.net/posts/20260510-schools-retreat/?utm_source=mastodon&utm_medium=social
A multi-agent LLM where each agent learns when to defer to a human, trained with GRPO on a cost-aware reward. Each defer event becomes SFT data, so the model gradually absorbs the human's expertise. Tunable cost knob trades accuracy against human-call budget at deployment, no retraining.
#ICLR #HumanInTheLoop #AgenticSystems #Metacognition #RL #AI