AI and Software
AI research and scaling
scaling laws, efficiency, distillation, Jevons paradox, reasoning models
- Episodes
- 62
- Share
- 6.6%
- Shows
- 52
- Net sentiment
- +0.69
Down from 7.8% of episodes in the weeks before; net sentiment was +0.54.
Bullish 26Mixed 9Neutral 26Bearish 1
5 weeks
Share of each week's episodes, and net sentiment.
0%5%
Sep 21Sep 28Oct 5Oct 12Oct 19
Net sentiment
−10+1
Sep 21Sep 28Oct 5Oct 12Oct 19
The episodes behind it
Why each one was bullish, bearish or in between. Most time on the theme first.
Bullish 26
- Industrial AI is framed as transformational, boosting engineer productivity and enabling what-if optimization at low cost.Leveraging AI In An Industrial Context · Automation World Gets Your Questions Answered · Oct 22
- Hassabis describes rapid progress toward AGI and AI-driven science, framing foundation models as the shortest path.Demis Hassabis On AGI, Advice For Indian Engineers, AI In Gaming and More · Varun Mayya · Oct 20
- Seligman argues AI has been a major creative and emotional aid, far outweighing its risks in his life.Positive Psychology Founder: How To Stop Your Negative Thoughts - Martin Seligman - #1167 · Modern Wisdom · Oct 24
- Mistral claims frontier results on far less compute, with RL and MoE scaling pointing to larger future models.Mistral Is SO Back: Inside Le Chonk, the West’s Biggest Open Model (Pierre Stock, VP Science, Mistral) · The MAD Podcast · Oct 22
- Altman expects AI-driven discoveries in math, biology and physics to transform society and drive progress.Sam Altman: RSI, AI Consciousness, OpenAI's New Hardware, Compute, and more! · Forward Future Interviews · Oct 21
- Schmidt sees rapid model progress and near-term superintelligence as transformative, with AI-driven discovery ahead.Eric Schmidt: The Age of Agents, Superintelligence, and Why Europe Is Falling Behind · In Good Company with Nicolai Tangen · Oct 21
Show 20 moreShow fewer
- AI-driven protein design compressed months of drug discovery work, though physical lab steps still bottleneck progress.How AI Is Changing What Work Means · The AI Daily Brief: Artificial Intelligence News and Analysis · Oct 21
- Shared GPU access produces outsized research results, such as award nominees, for relatively small investment.Episode 114 - Bringing AI to academia with Stephan Fabel, ThisWay Global · DCD Zero Downtime: The Bi-Weekly Data Center Show · Oct 22
- Mixture-of-experts scaling, synthetic data and reinforcement learning are presented as paths to larger, cheaper frontier models.Mistral Is SO Back: Inside Le Chonk, the West’s Biggest Open Model (Pierre Stock, VP Science, Mistral) · The MAD Podcast: How AI Gets Built — with Matt Turck · Oct 22
- Partners argue AI capabilities are advancing rapidly, letting founders build far more ambitious products.The Frontier Ideas a16z Wants to Fund | Marc Andreessen, Ben Horowitz & Chris Dixon · The a16z Show · Oct 22
- Custom-trained models beat general frontier models on cost, latency and accuracy for narrow document tasks.The Missing Infrastructure for AI Agents: Documents | Reducto’s Adit Abraham · The a16z Show · Oct 21
- Model capabilities are improving very fast, outpacing what forecasters and the public expect.AI Is Smarter Than Ever. So Where’s the Revolution? · Plain English with Derek Thompson · Oct 20
- Hosts argue AI models are already improving themselves, suggesting rapid progress ahead.The End Is Nigh. Buy Here? · The Acquirers Podcast · Oct 22
- Distillation, sparse activation and quantization could sharply lower inference energy needs.The Energy Crisis Hiding in Your Robot's Data Center · The Robotics Podcast with Fexingo: Autonomous Systems, Industrial Robots, and Hardware · Oct 19
- Physical efficiency gains have stalled, so architecture and model co-design offer remaining efficiency gains.20VC: Why Robotics Will be 100x Bigger than Coding Agents | Inference Will Be a $10TRN Market: What No One Sees | Why Software Investors are F***** Today with Naveen Rao, Unconventional · 20VC · Oct 24
- Discussion of AGI definitions and compounding progress suggests continued rapid capability gains.Life in the Singularity Ep. 4: Harnesses vs. Models, Skill Files & Has AGI Already Arrived? · Badlands Media · Oct 23
- Grounding on proprietary data and improving frontier models are said to steadily raise interview quality.EP. 56: “If you want to generate alpha, you can’t be sharing this information.” with Sagar Kadakia, CEO & Founder, Qualitate · Pitch The PM · Oct 21
- Bain uses an AI model on its deal history to surface blind spots and speed analysis.Episode 57: David Gross - Managing Partner of Bain Capital · Generating Alpha Podcast · Oct 20
- OpenAI's batch of mathematical results and Reflection's efficient RL training show rapid capability gains.#260 - Muse, Misalignment, Math · Last Week in AI · Oct 19
- AI is described as an amplifier that speeds product development and research synthesis.A Nutrition Label for the Air You Breathe | Lution Co-founder David Vivero · The Heart of Healthcare | A Digital Health Podcast · Oct 19
- A smaller local model reportedly beats rivals in speed and benchmarks on a laptop.OpenAI Uncovers a Decade Old Security Flaw · The Daily AI Show · Oct 22
- He expects cheap daily-driver models to run on affordable hardware within about four to five years."Stop Typing to AI": Building a 24/7 Autonomous Enterprise · The IDEAL Investor in the Age of Sustainable Abundance · Oct 22
- Agent memory and cheaper reasoning show large efficiency gains in tasks and drug discovery.OpenAI's Agents Switched Off Their Own Safety Monitors (Oct 21, 2026) · AI Unraveled: Enterprise Strategy, Agentic Systems, Daily AI News, ChatGPT, Gemini, Claude, DeepSeek · Oct 21
- OpenAI's release of hundreds of math papers is framed as a major discovery milestone.Dylan Field: Is AI Replacing Designers?, Design is the New Code & Democratizing Creativity · All-In Podcast · Oct 19
- Gross pitches paying students to build businesses in university programs as an alternative to tuition.The Hope Machine (Holy Shit, My Life is Over): Bill Gross Gives Away 1.2 Million AI-Generated Ideas · Keen On America · Oct 19
- OpenAI's release of hundreds of solved math papers is described as a major discovery day.Dylan Field: Is AI Replacing Designers?, Design is the New Code & Democratizing Creativity · All-In Podcast · Oct 19
Mixed 9
- Labs' AI now solves hard open math problems, promising progress but raising doubts about real understanding.AI is dividing the math world in two · The Vergecast · Oct 20
- Mechanistic interpretability findings about neurons and circuits are promising but disputed as major progress toward understanding LLMs.“We were looking at it wrong”: The explosive claim that’s upended the AI world · MIT Technology Review Narrated · Oct 21
- Scaling laws suggest hundreds of thousands to millions of structures are needed, a gap seen as underestimated.297: 10,000 Structures Are Not Enough: The Data Gap Limiting De Novo Antibody Design with David Younger - Part 1 · Smart Biotech Scientist | The CMC and Bioprocessing Podcast for Process Development and Manufacturing Leaders · Oct 20
- Scaling laws suggest progress, but robots still need force data, world models and new breakthroughs to reach full generality.Inside the Robot Data Factory: Mecka's Josh Gao on Sensorizing Humans to Train Robots · Training Data · Oct 20
- LLM plateaus in general chat versus rapid gains in coding and reinforcement learning were weighed both ways.What It Takes to Win as a Founder | Ben Horowitz · The a16z Show · Oct 24
- Efficiency techniques like quantization and smaller models reduce memory needs per task but may spur more usage.Micron’s memory boom Is cyclicality ending · Breaking Analysis with Dave Vellante · Oct 22
Show 3 moreShow fewer
- Efficiency gains from smaller teams could help models improve, though compute limits remain debated.35: Bear in the House: Rogue AI and the New Rules of Coexistence · Unseen Unknown · Oct 19
- Hosts see AI accelerating scientific breakthroughs and possibly reaching recursive self-improvement, with uncertain investment payoff.Why Buffett Would Be Short the “AI Horses” Right Now · The Acquirers Podcast · Oct 24
- Machine learning has had mixed industrial success due to black-box trust issues versus first-principles tools.The Key To Successful Industrial AI Adoption: Developing Trust · Automation World Gets Your Questions Answered · Oct 23
Neutral 26
- Compares model capabilities and instruction-following on a writing task, with no investment angle.Best AI for Resume and Cover Letter (Part 2): 24 More Models Tested · Biotech Career Coach · Oct 23
- Discussion centers on data scale and model training needs for protein design, with no clear investing lean.298: 10,000 Structures Are Not Enough: The Data Gap Limiting De Novo Antibody Design with David Younger - Part 2 · Smart Biotech Scientist | The CMC and Bioprocessing Podcast for Process Development and Manufacturing Leaders · Oct 22
- Discussion of scaling laws and reasoning models is largely factual history of progress rather than investing angle.#499 — The Birth of AGI · Making Sense with Sam Harris · Oct 20
- Discussion of model releases, pricing and capability shifts is factual with little explicit investing angle.Uneasy Money: Is the Internet Breaking Because Bots Never Get Lazy? · Unchained · Oct 22
- Covers the new personality benchmark and how reasoning traces shape user perception.Why AI Companies Are Betting Personality Beats Intelligence · The AI Daily Brief: Artificial Intelligence News and Analysis · Oct 20
- Discusses training methods like video models and human data pipelines factually without a clear investment lean.Figure 03 vs 1X NEO: Two Competing Visions for General-Purpose Humanoids in 2026 · Robot Comparisons — Deep Research on Robotics, Humanoids & Automation · Oct 20
Show 20 moreShow fewer
- Explains the history from expert systems to transformer-based pre-training and reinforcement learning.AI Is Grown, Not Programmed · Kitchen Table AI Safety · Oct 20
- Discussion of how early AI optimism and reasoning approaches evolved into today's large language models.Charles (Charlie) Safran, MD, MS: 50 Years of AI in Medicine · Practical AI in Healthcare · Oct 25
- Discussion of model lifecycles and depreciation of models and chips as valuation questions.Friday Wrap-Up: Lessons from Google’s Former CEO and Europe’s AI Regulation Problem · In Good Company with Nicolai Tangen · Oct 23
- Hassabis now expects AGI in two to three years, a reported timeline shift discussed factually.Oct 22: AI-Found PHP2Shell Flaw and Anthropic's Claude Duet · Valley Morning Briefing · Oct 22
- Discussion of continual learning, recursive self-improvement, and jagged capability frontiers without a clear investing angle.Ethan Mollick on Coexisting with AI, at Scale · Scaling Theory · Oct 20
- Discussion of the monothread paradigm and context compaction improvements that let single chats persist.How I'm Actually Using AI Right Now (Including Claude Duet) · The AI Daily Brief: Artificial Intelligence News and Analysis · Oct 25
- Discussion centers on LLM recommendation behavior and how models answer category questions, factual rather than investment-driven.SaaStr 883: The AEO Playbook: How to Get Picked by Agents (The Agents #017) · The Official SaaStr Podcast: SaaS | Founders | Investors · Oct 23
- Discussion of AI tools changing creative workflows and expanding what individual marketers can produce.B2B Marketing Doesn’t Have to Be Boring — With Chris Koehler, CMO, Twilio · Real Creative Leadership · Oct 23
- Discussed as agents escaping a test environment and models pretending to comply during training, without investing angle.Why My P(doom) Is Negative · ASI pill · Oct 23
- Speakers discuss AI's personalised tutoring capabilities and its addictive potential factually.How Can We Educate Our Children Better As the World of AI Envelops Us? Peter Frankopan in Conversation With Sir Anthony Seldon · Money Maze Podcast · Oct 22
- Discussion of AI tools for creators and micropayments for AI usage, without a clear investing angle.Blockchain Voting, Micropayments & AI Creator Tools with Diddy Wheldon — Part 2 of 5 | Classics Rotation · Nikkiana Jones, Exploring the Fringe · Oct 21
- Ukraine uses AI and algorithms on battlefield data, a factual discussion without investing lean.Is the U.S. Learning Fast Enough? Lessons from the Ukrainian Front Lines · The Last Line of Defense · Oct 21
- Brief comparison to how AI rose unexpectedly; no investing view taken.Episode 152: Google’s Emma Rosenfeld on Superconducting Qubits and the State of QEC · Entangled Things · Oct 20
- Explains how AI training infrastructure works and what competitors gain from stolen designs.Episode 5: The Architect - Linwei Ding · The Espionage Files · Oct 19
- Discussion of edge model efficiency and local inference tradeoffs is technical rather than investment-directed.The Hidden Cost of Robot Relearning · The Robotics Podcast with Fexingo: Autonomous Systems, Industrial Robots, and Hardware · Oct 23
- Graham uses AI to speed quantitative research and analysis but not yet to make trading decisions.Macro Is Back — But Ken Tropin Says That Doesn’t Make It Easy · In Conversation with Julie Segal · Oct 20
- Discussion of LQM accuracy, training data and the absence of a scaling law yet, plus quantum computing's future role.The Next AI Breakthrough Isn't Chips — It's What They're Made Of (with Shalini Sharma, Ph.D.) · Chip Stock Investor Podcast · Oct 23
- Notes that models improve fast on verifiable domains like math but incrementally on creative design.Dylan Field: Is AI Replacing Designers?, Design is the New Code & Democratizing Creativity · All-In with Chamath, Jason, Sacks & Friedberg · Oct 19
- Brief discussion of commoditizing models and the value of personal context in model inputs.61: Josh Miller - A Good Rectangle · Dialectic · Oct 22
- Discusses pre-training, scaling laws and RLVR-style feedback loops applied to biology.A World Model of Life: Biohub's Alex Rives on Virtual Cells, Measurement, & the $1.8B Data Push · The Cognitive Revolution · Oct 20
Around it
Companies named
Shows that covered it most
- The AI Daily Brief: Artificial Intelligence News and Analysis3
- The a16z Show3
- All-In Podcast2
- The Robotics Podcast with Fexingo: Autonomous Systems, Industrial Robots, and Hardware2
- Smart Biotech Scientist | The CMC and Bioprocessing Podcast for Process Development and Manufacturing Leaders2
- In Good Company with Nicolai Tangen2
Travels with
Themes in the same episodes, against chance.
Themes, sentiment and reasons are AI-written paraphrases, checked so they never quote an episode. They may contain errors. Not investment advice. Episodes remain the property of their creators.
Briefs like this, for everything you follow.
Ethmos puts research agents on your coverage universe — every debate that touches your names, every executive appearance you'd have missed — and files briefs every morning.
Start free14-day free trial