Marco Herzog 
www.computerworld.com
A moderated list for AI posts. Insights, experience, and AI news. Using 186 filters in total. No generated images etc.
Feed on BlueskyFeeds Stats
- 💙 Liked by 5 users
- 📅 Updated 10 months ago
- ⚙️ Provider skyfeed.me
AI feed ✨ Likes over time
Like count prediction
The feed AI feed ✨ has not gained any likes in the last month.
Feed Preview for AI feed ✨
Claude Code works better in poor internet than OpenAI Codex CLI. The former can work on a plane with meh wifi (maybe not perfectly), the latter couldn't even get started.
0
0
1
The invisible backbone of the UAE’s AI superpower dreams restofworld.org/2025/khaz… #AI #DataCenters #UAE
0
0
0
Moondream 3
A 9B param, 2B active MoE vision hybrid reasoning vision language model that supports both reasoning and non-reasoning mode. It focus on visually grounded reasoning, where the model references objects and spatial positions in the image while doing said reasoning,
1
1
4
@yannlecun and his collaborators (Hai Huang and Randall Balestriero) have released their first iteration of Joint Embedding Predictive Architectures (JEPAs) for LLMs.
"Learning by latent space prediction has revolutionized vision models, and it will revolutionize LLMs!, which they termed as
1
2
13
For those of you who are developing an AI-based finance research app: I’m not sure this is useful to evaluate your apps, but they have released the paper, the code, and the dataset.
ByteDance in collab with Columbia Business School released the FinSearchComp
1
0
4
xAI's Grok 4 Fast with a 2M context window.
FREE on OpenRouter, Vercel, Cline, OpenCode, Orchids App, and more for a limited time.
Grok 4 Fast is also generally available via the xAI API, with pricing starting at $0.20 / 1M input tokens and $0.50 / 1M output tokens.
x.ai/news/grok-4-...
0
0
3
Dr. Damien P. Williams Wants A Better Sociotechnical Paradigm
@wolvendamien.bsky.social
about 3 hours agoI have been saying this for 3 years now; it is evident from even a theorerical understanding of these systems' operations. But ppl keep claiming it's not true or not important, or could be patched over, & so I keep hoping each time an "AI" company CEO comes out & admits will be the one that sticks
Prof Mike Yearworth@mikeyearworth.bsky.social
"In a landmark study, OpenAI researchers reveal that large language models will always produce plausible but false outputs, even with perfect data, due to fundamental statistical and computational limits." www.computerworld.com/article/4059...
2
56
138
With the new Grok 4 Fast, the price/performance curve for AI shifted again. I updated my chart to reflect this change. I have to update it every couple weeks now.
I also think the GPQA Diamond benchmark is likely maxed out (the tests themselves have errors, making it impossible to get to 100%),
2
2
34
how much did google pay to slip an entire magazine of corporate propaganda into this weekend's @irishtimes.com newspaper disguised as public communication about AI
3
30
96
Great point.
Benchmarking has become the name of the game in machine learning.
You can effectively add to or subtract from the definition of intelligence by introducing a new benchmark or critiquing an existing one, and convince a significant portion of the community that you have a valid point.
Brad Aimone@jbimaknee.bsky.social
Incidentally, my personal favorite interpretation of advancements in computer science is that most of the progress we've made has been in defining the problems themselves. Many successes correlate to clearer definitions of what "good" means or abstracting the task away from weak definitions.
1
1
13
Some TPOT and e/acc folks have started popping in again lately. Hi! Your unhinged posts and chatbot romances are essential to our healthy AI ecosystem, so please check out my old post on constructing your feed to make bsky enjoyable.
Naomi Saphra@nsaphra.bsky.social
I wrote something up for AI people who want to get into bluesky and either couldn't assemble an exciting feed or gave up doomscrolling when their Following feed switched to talking politics 24/7.
nsaphra.net
4
6
57
Frankly going to implore progressives to follow better what is happening in China tech. It's not a VC bubble. It's a dozen top labs, it's most large co getting heavily into AI (at a pretraining level) and robotics, it's a deep bet on the next industrial revolution.
3
8
44
>80% of the issues you see with models is just training. remainder is internal token flow/squashing/entropy. there is no black box if you look inside.
Ethan Mollick@emollick.bsky.social
The jaggedness of AI remains even as models have rapidly come to exceed human abilities in many of the hardest timed math & science contests. Yet there is much less progress on good puns. True AGI would be figure out our limits in more than calculus (sorry, but also seriously).
0
0
7
AI Sign Language Translation: Ready for Education? - Artificial intelligence nationalcentreforai.jisci… #AI #accessibility
0
0
0
Oxford becomes first UK university to offer ChatGPT Edu to all staff and students www.ox.ac.uk/news/2025-09… #AI #education
0
1
0
Core reasoning model pieces rn
1. Reasoning tokens -- thinking
2. Search -- acquiring relevant information
3. Code -- building
Most other tools are just a form of code or search. All of these relate to in-context learning.
0
2
11
Two posts here that could have a fruitful conversation are, on the one hand, @ai4geo.bsky.social 's thread on ACI: defining goals for AI that are explicitly not human-shaped. +
Sam Barrett, PhD@ai4geo.bsky.social
Many people find ways to compliment and extend their cognitive abilities with AI (and many unfortunately don't extend but replace). In that context, what we have now and what would continue to be valuable in the future is ACI: Artificial Complimentary Intelligence...
2
0
15
Beetje technisch, maar de energiezuinige AI chips van Euclyd zijn wel heel interessant. Uit Eindhoven.👍
IO+@nl.ioplus.nl
Euclyd komt met exascale tokenfabriek met minimaal stroomverbruik https://ioplus.nl/nl/posts/euclyd-komt-met-exascale-tokenfabriek-met-minimaal-stroomverbruik
ioplus.nl
0
0
0
Job Prediction of Next Decade
Is Your Job Safe?
Spot the trends:
- Declining: Cashiers, Telemarketers
- Booming: AI Engineers, Data Scientists
- Rising: Telehealth, E-commerce
Stay ahead. Future-proof your career!
For more, see the infographic.
#careers #motivation #productivity
0
0
0