large language model

also referred to as: large language models · llm

46 statements across 38 episodes · 16 bullish · 15 bearish · 33 people on the record · first statement Apr 25, 2023 by Avlok Kohli · across every show →

Everything said about large language model, oldest first

Apr 25, 2023 positive
Disclosure
Kohli: Large language models have automated a good portion of AngelList operations
“A lot of the large language models and their advancements there have had a major impact on AngelList, just our own internal operations, right? It's been huge and we've been tinkering on it for many months now, and it's actually automated a good portion of the …”
Avlok Kohli Apr 25, 2023 ▶ 27:54 AngelList CEO Avlok Kohli: The Funding Market Today; Last-Mile Delivery; Competing w/ Carta | E1005 · 20VC with Harry Stebbings
May 9, 2023 positive
Assertion Not checkable as stated
Kohli: Large language models automated a good portion of AngelList operations
“And a lot of the large language models and their advancements there have had a major impact on AngelList, just our own internal operations, right? It's been huge. And we've been tinkering on it for many months now, and it's actually automated a good portion of…”
Avlok Kohli May 9, 2023 ▶ 20:35 How Dire is the Startup Funding Market Today? | AngelList CEO Avlok Kohli · 20VC with Harry Stebbings
Jun 19, 2023 positive
Opinion
Lebrun: AI application startups are not merely thin layers over LLMs
“So I think it's really not fair to think that any AI application on top of LLM is just a thin layer.”
Alex Lebrun Jun 19, 2023 ▶ 13:21 Alex Lebrun: Why the EU's AI Regulation is a Disaster; How Zuck Prepares for Meetings | E1027 · 20VC with Harry Stebbings
Jun 19, 2023 negative
Insight
Lebrun: Training LLMs on Curated Data Does Not Guarantee Trustworthy Outputs
“Even if you feed an LLM with curated data, you don't, it does, it's not guaranteed that the output will be will be good, will be perfect, that you can trust the output. So, feeding an LLM with trusted data doesn't make the output trustable because of how LLM w…”
Alex Lebrun Jun 19, 2023 ▶ 29:39 Alex Lebrun: Why the EU's AI Regulation is a Disaster; How Zuck Prepares for Meetings | E1027 · 20VC with Harry Stebbings
Jun 30, 2023 negative
Assertion Supported
Douwe Kiela: Inability to delete LLM data creates GDPR compliance issues
“So we can't really remove information from them, which is kind of tricky from a GDPR perspective.”
Douwe Kiela Jun 30, 2023 ▶ 0:13 Douwe Kiela: Why Data Size Matters More Than Model Size; Why Open Source Isn't Going to Win | E1032 · 20VC with Harry Stebbings
Jun 30, 2023 positive
Insight
Douwe Kiela: Decoupling LLM memory reduces hallucinations and enables data deletion
“You decouple the memory from the generative capacity of the large language model, and this allows you to ground the generations from the language model into things you retrieved. In your memory, essentially. So you get much less hallucination. You get attribut…”
Douwe Kiela Jun 30, 2023 ▶ 9:21 Douwe Kiela: Why Data Size Matters More Than Model Size; Why Open Source Isn't Going to Win | E1032 · 20VC with Harry Stebbings
Jul 24, 2023 positive
Disclosure
Adam Mosseri Uses LLMs to Summarize Performance Feedback and Edit Writing
“I write pretty detailed reviews for my reports every six months and so I don't have AI write it, but I have it take a first pass at summarizing all the feedback to make sure I didn't miss anything. I will sometimes Give it a piece. I do a lot of writing at wor…”
Adam Mosseri Jul 24, 2023 ▶ 1:18:40 Adam Mosseri: Why Video Is Bad For Business; Does Feed Matter Anymore? | E1039 · 20VC with Harry Stebbings
Aug 7, 2023 neutral
Prediction Not checkable as stated
Khosla: Only two or three major LLM providers will emerge
“There'll be two or three major providers of large language models.”
Vinod Khosla Aug 7, 2023 ▶ 15:49 Vinod Khosla: How AI Impacts Healthcare, Energy, Geo-Politics, Media and Climate Change | E1045 · 20VC with Harry Stebbings
Aug 18, 2023 negative
Opinion
Socher: AI safety guardrails have overshot by censoring benign creative writing
“And I think some cases we even overshot a little bit. You know, there are some folks who say, oh, you cannot tell any dirty jokes anymore, or like a murder mystery is like unethical to write.”
Richard Socher Aug 18, 2023 ▶ 23:51 Richard Socher: The 3 Biggest Barriers to Building AGI; AI Startups vs Incumbents | E1050 · 20VC with Harry Stebbings
Aug 18, 2023 positive
Insight
LLMs allow small startups to rapidly build 80% AI product MVPs
“Because of these large language models and foundational models we can have a general sense of understanding natural language, and you can, as a small startup nowadays, build in like, 80% solution very quickly. And then you layer more and more specific data for…”
Richard Socher Aug 18, 2023 ▶ 18:41 Richard Socher: The 3 Biggest Barriers to Building AGI; AI Startups vs Incumbents | E1050 · 20VC with Harry Stebbings
Aug 18, 2023 positive
Assertion Supported
Socher: Calling You.com a thin LLM wrapper is completely false
“But it's something that, you know, some VCs don't appreciate and understand the complexity of, and then they say, oh, like u.com is just a thin wrapper around a large language model, which is very far from the truth.”
Richard Socher Aug 18, 2023 ▶ 21:55 Richard Socher: The 3 Biggest Barriers to Building AGI; AI Startups vs Incumbents | E1050 · 20VC with Harry Stebbings
Aug 18, 2023 neutral
Disclosure
Socher: You.com displays live stock tickers instead of LLM-generated estimates
“Like new.com has, for instance, when you ask for a stock price, we don't just make up a bunch of numbers like an LL would. We actually show you a stock ticker and then you can see the actual price, right?”
Richard Socher Aug 18, 2023 ▶ 5:26 Richard Socher: The 3 Biggest Barriers to Building AGI; AI Startups vs Incumbents | E1050 · 20VC with Harry Stebbings
Aug 25, 2023 neutral
Insight
Howie Liu: Enterprise AI Adoption Remains in an Early Education Phase
“And there's also just a you know, we're still in the education phase where I think every enterprise is, is getting savvy as we speak or trying to get savvy as we speak in terms of what is, you know, what is an LLM? You know, what are these basic primitives, ri…”
Howie Liu Aug 25, 2023 ▶ 10:08 Howie Liu: Decoding Airtable's $11B Valuation; The Impending AI Revolution in Enterprise | E1053 · 20VC with Harry Stebbings
Sep 27, 2023 bearish
Prediction Not checkable as stated
Lanng: Most application-based software brands will struggle due to AI
“I think actually that one of the biggest impacts of AI and LLM is that most application-based brands are going to be really struggling because the brand is the UX.”
Christian Lanng Sep 27, 2023 ▶ 1:07:06 Christian Lanng: "How Being a Founder Almost Killed Me" | E1065 · 20VC with Harry Stebbings
Dec 20, 2023 positive
Prediction Not checkable as stated
Cohen: LinkedIn is using LLMs to create interfaces that morph to users
“One thing that we're already underway for us is to really take away the complexity and solve it with the idea of bringing in more of this new large language models, and in general, this new brain that can exceed on top of it. So instead of adding more features…”
Tomer Cohen Dec 20, 2023 ▶ 52:07 Roundtable #7: Spotify, Adobe and Linkedin on How AI Changes The Future of Product & Design | E1097 · 20VC with Harry Stebbings
Feb 2, 2024 negative
Prediction Not checkable as stated
Will Wu: Traditional screen user interfaces will not be replaced by AI
“I don't think that the previous human computer interaction paradigms are going away. For instance, look at, like, the command line interface. That was one of the original ways to interact with computers. Like, people are still using command line interfaces tod…”
Will Wu Feb 2, 2024 ▶ 47:04 Will Wu: Top Five Product Lessons from Creating Snapchat "Discover" and "Chat" | E1111 · 20VC with Harry Stebbings
Apr 8, 2024 negative
Assertion Supported
Lehmann: High LLM compute costs prevent killer apps beyond chatbots
“I believe that one of the overlooked areas of the current state of AI and the large language models that we're using is that they're still very expensive. And as a result of that, they're dramatically more expensive than a Google search. And as a result of tha…”
Bastian Lehmann Apr 8, 2024 ▶ 49:08 Bastian Lehmann: How the Uber Deal Went Down and How a $2.65BN Deal Turned into $5BN | E1137 · 20VC with Harry Stebbings
Jul 19, 2024 neutral
Prediction Not checkable as stated
Cameron Adams: AI innovation speed will slow down within five years
“In five years time, I would say that AI innovation will have slowed by then, and a lot more will be standardized. Social patterns will be established of like how we interact with AI. The possibilities of what LLM can do are very well understood and known, and …”
Cameron Adams Jul 19, 2024 ▶ 17:24 Cameron Adams: How Canva Builds Products: Lessons Learned, What Works? What Flopped? | E1179 · 20VC with Harry Stebbings
Jul 31, 2024 bullish
Insight
Mollick: AI value accrues to skilled integrators, not compute providers
“The real value of the steam engine came from having skilled artisans in your factory who said, I've got this thing that can make power go back and forth. How do I create the gearing to connect that to my, you know, my spinning Jenny, my ammunition manufacturin…”
Ethan Mollick Jul 31, 2024 ▶ 13:25 Ethan Mollick: Why OpenAl Abandons Products, The Biggest Opportunities They Have Not Taken | E1184 · 20VC with Harry Stebbings
Aug 26, 2024
Insight
Khan: Large language models are entry points like web browsers, not ultimate value
“LLM is just the browser. In my mind, there's like LLM is not different than the browser. Like obviously it's very different, but it's the entry point to what you want to do.”
Imran Khan Aug 26, 2024 ▶ 47:29 Imran Khan: Why the IPO Market is Not Closed & Lessons From Taking Snap & Alibaba Public | E1194 · 20VC with Harry Stebbings
Oct 14, 2024 bearish
Prediction Not checkable as stated
Frankel: LLM tools will commoditize, letting SaaS companies buy AI infrastructure cheaply
“I think the tools, I actually think some of the LLM and the, you know, I think some of the tools are actually very commoditized. So I think that the vertical SaaS or horizontal SaaS or data owners We'll be able to access the underlying tools reasonably cheaply…”
David Frankel Oct 14, 2024 ▶ 1:08:39 David Frankel, MP @Founder Collective: Investing Lessons from Seeding Coupang, Pillpack & Suno|E1214 · 20VC with Harry Stebbings
Dec 16, 2024 bullish
Prediction Not checkable as stated
Hoffman: LLM scaling continues and larger models will show new magic
“We're not reaching the upper end of LLM. The next large LLM that's training with a larger computer will still have new magic in it. The scale game is still playing.”
Reid Hoffman Dec 16, 2024 ▶ 0:17 Reid Hoffman, LinkedIn & Paypal Founder: Trump Administration, Elon Musk and DOGE | E1239 · 20VC with Harry Stebbings
Dec 18, 2024 neutral
Insight
Daniel Dines defines enterprise AGI as predictable 120 IQ multitask capability
“AGI for enterprises would be when I have an LLM that has the capabilities of a guy with an average IQ like a 120 points. But predictably. Not a 180 in some fancy math jobs and 60 into, you know, other type of job. Predictably. This is AGI.”
Daniel Dines Dec 18, 2024 ▶ 36:50 Daniel Dines, UiPath CEO & Founder: Why Agents Do Not Mean RPA is F*** | E1240 · 20VC with Harry Stebbings
Jan 15, 2025 negative
Opinion
Riparbelli: Claiming AI prompt tools will easily kill software incumbents is naive
“And I think that's like a really, really naive view of what it's like to build a business. Technology and lines of code is super important, but as you know, right, you've built your company yourself. It's a, it goes way deeper than that. And I think it really,…”
Victor Riparbelli Jan 15, 2025 ▶ 22:17 Victor Riparbelli, CEO @Synthesia: OpenAI vs Anthropic vs X.ai - Who Wins and Why | E1246 · 20VC with Harry Stebbings
Jan 29, 2025
Insight
Ross: Generative AI Is Fundamentally Different From the Internet
“Is AI the next internet? And I'm like, absolutely not, because the internet is an information age technology. It's about duplicating data with high fidelity and distributing it. It's what telephone does. It's what internet does. It's what the printing press di…”
Jonathan Ross Jan 29, 2025 ▶ 34:58 Jonathan Ross: DeepSeek Special - How Should OpenAI and the US Government Respond | E1253 · 20VC with Harry Stebbings
Jan 29, 2025
Assertion Supported
Ross: Google announced first zero-day exploit found by an LLM
“Google just announced recently the first zero day exploit found by an LLM that was previously unknown.”
Jonathan Ross Jan 29, 2025 ▶ 54:45 Jonathan Ross: DeepSeek Special - How Should OpenAI and the US Government Respond | E1253 · 20VC with Harry Stebbings
Feb 17, 2025 positive
Insight
Ross: LLM-generated synthetic data is better for model training
“You could have an LLM generate synthetic data, and when it generates the synthetic data, the data is better. You then train on that synthetic data.”
Jonathan Ross Feb 17, 2025 ▶ 3:09 Jonathan Ross, Founder & CEO @ Groq: NVIDIA vs Groq - The Future of Training vs Inference | E1260 · 20VC with Harry Stebbings
Mar 5, 2025
Insight
Osika: Putting LLMs inside a for loop enables autonomous AI agents
“Hook them up and make you basically put the large language model in a for loop and then you can have it do a lot of agentic things.”
Anton Osika Mar 5, 2025 ▶ 7:15 Anton Osika, Co-Founder and CEO @ Lovable: Hitting 85% Day 30 Retention - Better than ChatGPT · 20VC with Harry Stebbings
Mar 21, 2025
Assertion Supported
Elias Torres: Code to build an LLM fits on one page
“The code fits on a, like, a page on the screen.”
Elias Torres Mar 21, 2025 ▶ 4:36 Elias Torres, Co-Founder and CEO @ Agency: What No One Tells You About Selling Your Company · 20VC with Harry Stebbings
Apr 10, 2025 negative
Assertion Supported
Hulme: High UK energy costs prevent competitive AI model training
“We have the, probably the highest electricity or energy costs in the whole of the Western world. In the UK. That just does not enable you to do a great job of this. It doesn't even enable you to do a great job of training foundation models. Like, if the blende…”
Tom Hulme Apr 10, 2025 ▶ 24:38 Tom Hulme & Stan Boland: Lessons from Jensen Huang & How to Fix the UK Tech Ecosystem · 20VC with Harry Stebbings
Apr 18, 2025 positive
Insight
Socher: Enterprise Is the Killer Application for Large Language Models
“We realized eventually the killer app for large language models and complex answers is an enterprise.”
Richard Socher Apr 18, 2025 ▶ 5:44 20VC: Foundation Models: Who Wins & Who Loses | How Economies and Labour Markets Need to Change in a World of AI | China vs the US in an AI Race: What You Need to Know | Rich Socher, Founder @ You.com
Jun 12, 2025 negative
Insight
O'Driscoll: AI developers don't need to license more than two or three news outlets
“You could imagine an LLM saying, I need to get modern news from AP and one national newspaper, but I sure as hell don't need six. Because if you think back... By the time you got to the third, you're like, I know already, 90% of the content is repetitive”
Rory O'Driscoll Jun 12, 2025 ▶ 1:18:43 Elon’s Empire: SpaceX, Tesla, Neuralink After the Storm & Anduril’s $2.6BN Power Move · 20VC with Harry Stebbings
Jul 11, 2025 bearish
Prediction Open · timeframe Dec 2027
Flanagan: 95% of B2B buying journeys will start in LLMs by 2027
“Google search is one of the best channels we had in B to B. 80% of all purchases started on Google search. It's predicted by 2027, 95% will start in LLM.”
Kieran Flanagan Jul 11, 2025 ▶ 1:04:14 20Growth: The Death of Growth Teams? | How Hubspot Use AI to Triple Email Conversion | The Future of AI SEO | Why Prompt Engineering is the New Coding | What Every CMO Needs to Know About AI in 2025
Jul 11, 2025 bullish
Opinion
Flanagan: Memory is an LLM's most defensible feature
“I think it's the most defensible part of an LLM.”
Kieran Flanagan Jul 11, 2025 ▶ 32:58 20Growth: The Death of Growth Teams? | How Hubspot Use AI to Triple Email Conversion | The Future of AI SEO | Why Prompt Engineering is the New Coding | What Every CMO Needs to Know About AI in 2025
Sep 1, 2025
Assertion Partly supported
Frosst: Fewer than 20 companies globally build foundational large language models
“Maybe there's like 10 companies in the world that are building, like, large language models. In the, yeah, maybe the, in, maybe there's like 15 in, I don't know, we'll have to figure out how, there's a few that have popped up recently. But there's some number,…”
Nick Frosst Sep 1, 2025 ▶ 3:41 Cohere Founder, Nick Frosst: How To Compete with OpenAI & Anthropic, and Sam Altman’s AI Disservice · 20VC with Harry Stebbings
Sep 1, 2025 neutral
Assertion Supported
Frosst: Large language models have made zero independent scientific or intellectual breakthroughs
“There has been no independent breakthrough that an LLM has made. Right. There has been no, like nobody has seen, nobody asked in LLM, Hey, solve this problem. No one's solved before and get the answer. The breakthroughs are still people.”
Nick Frosst Sep 1, 2025 ▶ 26:39 Cohere Founder, Nick Frosst: How To Compete with OpenAI & Anthropic, and Sam Altman’s AI Disservice · 20VC with Harry Stebbings
Sep 11, 2025 positive
Prediction Open · timeframe Sep 2030
O'Driscoll: Data Labeling Firms Will Sell Legally Compliant Book Datasets for AI
“I assume that there will be a much more efficient way than that of doing it. You're exactly right. I'm sure, for example, that there will be a corpus available of a purchased copy of every, you know, I'm sure that Scale some one of these label guys that says w…”
Rory O'Driscoll Sep 11, 2025 ▶ 44:17 OpenAI’s $10BN Secondary Sale, Ramp Hits $1BN ARR & Brex Hits $700M · 20VC with Harry Stebbings
Oct 13, 2025 positive
Assertion Not checkable as stated
Cannon-Brookes: Simultaneous global release of LLMs is unprecedented in tech history
“Large language models have been dropped on the whole world at the same time. It's like someone just went, you can all have these for free, go. Which is something I don't believe we've seen in technology before.”
Mike Cannon-Brookes Oct 13, 2025 ▶ 39:00 Atlassian CEO, Mike Cannon-Brookes on Why Everything is Overvalued & Are We in an AI Bubble · 20VC with Harry Stebbings
Nov 3, 2025
Assertion Supported
Pineau: Meta's 2019 RoBERTa model hit 20M monthly downloads during LLM hype
“We were in, in the frenzy of large language models, and I pulled the stats on, you know, most downloaded models of last month. We had a model like Roberta from 2019, small language model, was getting twenty million downloads a month.”
Joelle Pineau Nov 3, 2025 ▶ 57:31 Cohere's Chief AI Officer, Joelle Pineau: Why Scaling Laws Will Continue & Future of Synthetic Data · 20VC with Harry Stebbings
Jan 15, 2026 negative
Assertion Supported
Lemkin: The shelf life of a state-of-the-art LLM is under 100 days
“There is a very simple bear case that it goes almost to zero which is that shelf life of an LLM is less than a hundred days.”
Jason Lemkin Jan 15, 2026 ▶ 19:21 Anthropic’s $10B Raise | a16z’s $15B Fund: Is the Middle Dead in VC? | How OpenAI Could Go to Zero? · 20VC with Harry Stebbings
Mar 14, 2026 bearish
Prediction Not checkable as stated
Verna: LLMs will soon become cheap commodities like cloud access
“Because LLM is gonna be like a commoditized feature Very soon. And it doesn't feel so right now, but it will be like access to internet or access to the cloud. It's commoditized. It's cheap.”
Elena Verna Mar 14, 2026 ▶ 35:40 Elena Verna: How Lovable Launches Product & Hacks Social to Go Viral · 20VC with Harry Stebbings
Apr 27, 2026 negative
Insight
Adam Foroughi: Custom Recommendation Models Outperform Frontier LLMs for Ad Delivery
“You can't just go defer to the large language model and say, hey, based on what you know about this user and the data I have available, what's the next ad to see? That wouldn't work as well as a custom model built for this purpose.”
Adam Foroughi Apr 27, 2026 ▶ 33:07 AppLovin CEO: Why Founders Shouldn't Angel Invest & Why the Best Don't Need Mentorship · 20VC with Harry Stebbings
Jun 25, 2026
Insight
O'Driscoll: Converting medical AI discoveries into commercial value takes 10 to 15 years
“The lag time between I've had amazing invention and revenue in the kind of LL, core LLM space is a year, or two years, or three years. The lag time if you have a medical invention, an idea, and turning that into real meaningful economic value is 10 or 15 years…”
Rory O'Driscoll Jun 25, 2026 ▶ 7:15 DeepSeek Raises at $50B | The Rise of Open Source vs OpenAI & Anthropic | OpenAI Builds Own Chip · 20VC with Harry Stebbings
Jul 13, 2026 bearish
Prediction Not checkable as stated
Abrahami: LLMs will not replace white-collar jobs because they lack reasoning
“I would not be That concern with LLM is replacing every kind of a white color jobs. That's not going to happen. It's not that smart. You keep making silly mistakes. It's not very good. Actually, it's not very good at reasoning. It's very good at collecting dat…”
Avishai Abrahami Jul 13, 2026 ▶ 42:54 Wix Founder: Will Base44 Win the Vibe-Coding Wars? | The Truth About the Economics of Vibe-Coding · 20VC with Harry Stebbings
Jul 31, 2026 positive
Insight
Park: LLMs can simulate realistic human behavior across domains
“One of the early observations that we made was that these models are trained on so much of human behavior data, sentiment data that were expressed on the web. So if you poke at them sort of at the right angle, you could actually extract a lot of realistic huma…”
Joon Sung Park Jul 31, 2026 ▶ 2:03 The AI Company Simulating the Entire Economy | Simile Co-founder & CEO, Joon Sung Park
Aug 29, 2026 neutral
Assertion Not checkable as stated
Reyes: True continuous learning within a closed-loop LLM does not exist
“There was an idea of continuous learning from over the last couple of years that said that you would have a, like a literal LLM like model where all of the learning happens internal to this closed loop system. That technology has not been developed. It doesn't…”
Eno Reyes Aug 29, 2026 ▶ 32:29 Should American Enterprises Work With Open-Source Chinese Models? | Only 10% of Neo-labs survive
Made with StarZero

Turn any episode into a week of clips.

This entire site, over 1,200 episodes transcribed, diarized, checked and made playable, runs on the StarZero media pipeline. Drop in your own episode and the podcast clipper finds the moments worth sharing, cuts them, captions them, and reframes them for every feed.