#scalinglaws — Public Fediverse posts

“Why does every mammal get one billion heartbeats in its lifetime?”

I’ve just watched one of the most fascinating episodes of Veritasium!

Approximate animals as spheres and plot empirical data to uncover patterns, that’s physics. Explore power laws and fractals in nature, that’s mathematics. Explain why Tusko the elephant died after receiving an LSD drug dose extrapolated from smaller animals using linear proportions, and why humans experience many more heartbeats than most mammals, that’s biology. The beauty of interdisciplinarity!

Now I’m curious to read more, starting with Scale by Geoffrey West. West also co-edited Scaling in Biology and co-developed the WBE theory, which attempts to explain Kleiber’s law: the relationship between an animal’s metabolic rate and its body mass.

There’s a reason this video has already attracted 3.3 million views in only two days!

https://youtu.be/tL9Lw250spc

#Science #Physics #Mathematics #Biology #Fractals #ScalingLaws #Scale

@epochai.bsky.social

We've added narrations to our long-form content on the Epoch AI website, including reports, Gradient Updates, and topic overviews.

Look for the play button.

https://bsky.app/profile/epochai.bsky.social/post/3mmwufxh6i22j

#AI News #ScalingLaws

@emollick.bsky.social

The Second Scaling Law of AI remains undefeated.

If you want better hacking (or math, or science, or crossword puzzle solving) out of an LLM, just let it use more tokens. There doesn't seem to be any plateau so far in the new study by the UK's governmental AI Security Institute.

the reason this is dumb, imo, is it assumes the consumer will always want to buy/use yet another agent

https://bsky.app/profile/emollick.bsky.social/post/3mlu2ikfjls2s

#AI News #LLM #ScalingLaws

Robot magazziniere che improvvisa sul lavoro. Il Robot Generalista che Piega 200 Scatole Senza Errori: Arriva GEN-1 di Generalist AI. Generalist AI ha annunciato GEN-1, il primo modello robotico capace di superare il 99% di affidabilità nelle operazioni industriali. Offre una velocità di esecuzione tre volte superiore ai.

https://scienzamagia.eu/scienza-e-tecnologia/robot-magazziniere-che-improvvisa-sul-lavoro/

#Ambientiproduttivi #GeneralistAI #HarmonicReasoning #intelligenzaartificiale #Modelligeneralisti #reinforcementlearning #Scalinglaws #Taskindustriali

LLM ”emergence” is not true emergence. It’s only been called that because indoctrinated people wanted to believe that was what was happening.
And probably because it sounds good as a piece of magic to throw VC capital funders your whole business empire depends on.
https://youtube.com/watch?v=jXa8dHzgV8U
#ai #llm #hype #emergence #scalinglaws

@lawfare 🧵Scaling Laws — Lawfare podcast

“Mollick discusses the transformative potential of AI in various fields, particularly education and medicine, as well as the need for empirical research to understand AI's impact, the importance of adapting teaching methods, and the challenges of cognitive de-skilling.”

Includes extended discussion on law school teaching. The Apple podcast has a generated transcript.

#lawfare #ScalingLaws #podcast #AGI #AI #LLM #pedagogy

https://podcasts.apple.com/us/podcast/scaling-laws/id1607949880?i=1000715411530

Scaling Laws - I strongly recommend this #podcast series on #AI #LLM related #Legal topics, including Law #Education #ScalingLaws
https://mastodon.social/@lawfare/114874664114525328

🚀🤯 Dive deep into the riveting world of #SSMs vs. #Transformers where acronyms abound, and jargon reigns supreme! 💥 Witness the epic battle of "tokenization" and "attention" in this sensational snooze-fest. Spoiler: You won't care about "scaling laws" after reading this. 🤷‍♂️📉
https://goombalab.github.io/blog/2025/tradeoffs/ #tokenization #attention #scalinglaws #technews #HackerNews #ngated

@tsarnick

Reid Hoffman says we are not at the upper end of LLM capability: "we're still getting to the next set of exciting things... the next LLM will still have new magic in it... the scale game is still playing"

#ai #llm #scalinglaws

https://x.com/tsarnick/status/1868767587548774701?s=46

@wesroth.bsky.social
#ai #agi #scalinglaws

#OpenAI LEAK: "AI progress will HALT" | GPT-5 is Slowing Down!

https://youtu.be/duYRHq8eECo?feature=shared

@emollick.bsky.social

I wrote about scaling last week. Altman is arguing that it is, in fact, all you need.

#GenAI #OpenAI #ScalingLaws

https://www.oneusefulthing.org/p/scaling-the-state-of-play-in-ai

Microsoft CTO Kevin Scott thinks LLM “scaling laws” will hold despite criticism - Enlarge / Kevin Scott, CTO and EVP of AI at Microsoft speaks onstage du... - https://arstechnica.com/?p=2036166 #largelangaugemodels #machinelearning #aiscalinglaws #microsoftcto #scalinglaws #airesearch #kevinscott #microsoft #chatbots #biz⁢ #openai #gpt-4 #gpt-5 #ai

COMPLEXITY: Melanie Moses on Metabolic Scaling in Biology & Computation

I really enjoyed this podcast. It's neat how the amount of required infrastructure scales with size of a system. One thing I take away from this is that in some ways the scaling depends on the "technology" of the system. For example single membrane versus vasculature versus eusocial superorganism.

Episode webpage: https://podcasters.spotify.com/pod/show/md-ansar-md-ansar/episodes/Melanie-Moses-on-Metabolic-Scaling-in-Biology--Computation-eccmsq

#santafeinstitute
#complexity
#scalinglaws

What are larger models worse at?

The Inverse scaling competition was much discussed
for its novelty and the $100K prize

what did they find?

https://arxiv.org/abs/2306.09479
#NLProc #scaling #scalinglaws #pretraining #data #machinelearning