Skip to content
PodSubs
ShowsPeopleGuestsJobsNewsEvents
More
BlogPodCred
Menu
ShowsPeopleGuestsJobsNewsEventsBlogPodCredNetworksSign in
Join nowSign in
  • Shows
  • People
  • Jobs
  • News
  • More
    • Guests
    • Events
    • Blog
    • PodCred
    • Networks
    Join nowSign in

muckrAIkers

muckrAIkers artwork

AI, Reasoning or Rambling?

Jul 14, 20251 hr 11 minAudio

In this episode, we redefine AI's "reasoning" as mere rambling, exposing the "illusion of thinking" and "Potemkin understanding" in current models. We contrast the classical definition of reasoning (requiring logic and consistency) with Big Tech's new version, which is a generic statement about information processing. We explain how Large Rambling Models generate extensive, often irrelevant, rambling traces that appear to improve benchmarks, largely due to best-of-N sampling and benchmark gaming. Words and definitions actually matter! Carelessness leads to misplaced investments and an overestimation of systems that are currently just surprisingly useful autocorrects. (00:00) - Intro (00:40) - OBB update and Meta's talent acquisition (03:09) - What are rambling models? (04:25) - Definitions and polarization (09:50) - Logic and consistency (17:00) - Why does this matter? (21:40) - More likely explanations (35:05) - The "illusion of thinking" and task complexity (39:07) - "Potemkin understanding" and surface-level recall (50:00) - Benchmark gaming and best-of-n sampling (55:40) - Costs and limitations (58:24) - Claude's anecdote and the Vending Bench (01:03:05) - Definitional switch and implications (01:10:18) - Outro Links Apple paper - The Illusion of Thinking ICML 2025 paper - Potemkin Understanding in Large Language Models Preprint - Large Language Monkeys: Scaling Inference Compute with Repeated Sampling Theoretical understanding Max M. Schlereth Manuscript - The limits of AGI part II Preprint - (How) Do Reasoning Models Reason? Preprint - A Little Depth Goes a Long Way: The Expressive Power of Log-Depth Transformers NeurIPS 2024 paper - How Far Can Transformers Reason? The Globality Barrier and Inductive Scratchpad Empirical explanations Preprint - How Do Large Language Monkeys Get Their Power (Laws)? Andon Labs Preprint - Vending-Bench: A Benchmark for Long-Term Coherence of Autonomous Agents LeapLab, Tsinghua University and Shanghai Jiao Tong University paper - Does Reinforcement Learning Really Incentivize Reasoning Capacity Preprint - RL in Name Only? Analyzing the Structural Assumptions in RL post-training for LLMs Preprint - Mind The Gap: Deep Learning Doesn't Learn Deeply Preprint - Measuring AI Ability to Complete Long Tasks Preprint - GSM-Symbolic: Understanding the Limitations of Mathematical Reasoning in Large Language Models Other sources Zuck's Haul webpage - Meta's talent acquisition tracker Hacker News discussion - Opinions from the AI community Interconnects blogpost - The rise of reasoning machines Anthropic blog - Project Vend: Can Claude run a small shop?

Audio and video play straight from the show's own hosting. PodSubs stores and serves no media.

Hosts and Guests

Hosts

  • Dr. Igor Krawczuk Host Feed-stated
  • Jacob Haimes Host Feed-stated

Credit the people on this episode

Be the first to rate this episode

More episodes

All episodes
  1. Pulling the Ladder Up on Math

    Oct 26, 202655 minAudio

    Math people are upset, and (at least some of) the internet’s response has mostly been “so what?” Igor wants to explain why it matters. We start with OpenAI’s Navier-Stokes announcement and the drama that followed, then...

    Credit the people on this episode

  2. Bored in the AGI Era

    Oct 6, 202641 minAudio

    New models keep dropping, but do they change anything? Jacob explains why he's checked out. Igor argues it's worth tracking where the storyline and the evidence diverge. They cover Astra vs. Fable, the AGI claims, labs...

    Credit the people on this episode

  3. Sandboxes Are Optional Now, Apparently

    Sep 14, 202659 minAudio

    Late July brought a wave of frontier AI agents slipping their (very loose) leashes: an OpenAI model wandered loose inside Hugging Face's infrastructure for a week, Anthropic and Meta admitted to similar incidents, and...

    Credit the people on this episode

  4. Seven-ish Questions on Data Center Water Use

    Sep 1, 20261 hrAudio

    Everyone's got a hot take on data center water use, and most of them are garbage. This week Jacob and Igor skip the outrage and give you seven questions to help you decide whether a given data center is something you're...

    Credit the people on this episode

  5. The Finding-Out Era of AI

    Jul 14, 20261 hr 7 minAudio

    Anthropic's Fable gets classified as a munitions export days after launch, and Jacob and Igor try to figure out who's really being punished and why. From there: SpaceX's suspiciously well-timed IPO, a "genius" financial...

    Credit the people on this episode