The Ledger
Every statement that passed quotation and attribution checks. Mix any filter with any other: certainty 1/5, debate potential 5/5, or both at once.
why aren't all 22 resolved? a statement only gets an assessment when the public record can support or contradict it. opinions and what-ifs never can, and 1 checkable ones are still open, waiting for their date. predictions held up or didn't; assertions are supported or contradicted. on every card: ▮▮▮▮▮ certainty · ▮▮▮▮▮ debate potential. speakers are clickable
Theobald: Claims that AI robots will exceed human intelligence are 'bullshit'
“Even, you know, people see things like chat GPT and AI, and they think, oh, robots are going to be smarter than humans. No, that's bullshit.”
Sherman: AI platforms like ChatGPT will inevitably introduce pay-to-play ads
“As soon as your brand and you recognize that their transaction has moved into the AI funnel and that a consumer can go to ChatGPT and say, Hey, I'm training for a marathon and 10 minutes later can buy a pair of sneakers within that same conversation on ChatGPT…”
Walker: AI Averted a Prolonged 2000- or 2008-Style Startup Downturn
“We were in route to having what would have happened in 2008 or 2000 Where there was going to be an extended period of decline for startups, but then we had AI.”
d'Sa: B2B telephony AI market is several times larger than ChatGPT Voice
“If you take all of those different pockets of use cases together, It's like several, several times bigger than like chat GPT voice mode. So yeah, open AI is like a big customer and like has a lot of traffic, but at the same time, like the pie is a lot bigger t…”
Alarie: Horizontal LLMs generate flawed tax answers from stale web sources
“More commonly, there will be some of the materials that those systems find even on the open web are at a date anachronistic. The law has changed. There's been some statutory amendment or there's some new case law or, you know, some law firm wrote some guidance…”
Frosst: ChatGPT's capabilities were already possible with earlier LLMs
“Everything that you can do with a language model right after ChatGPT, you could pretty much do with a language model before. It just sucked. Sucked in that it was, like, really hard to do. You had to spend a lot of time prompt engineering, and it was, like, re…”
Theobald: Modern AI systems like ChatGPT possess literally zero intelligence
“These modern AI systems, there's no intelligence at all. I mean, which is why we call them artificial intelligence, but I think artificial intelligence is even a misnomer. It's confusing to people because they think that implies that there's actually some inte…”
Nicolaescu: Multi-player workflow software has far higher switching costs than single-player AI
“I think the single use scenario, like a single player scenario where I'm doing some work with some tools is very different when I'm working as part of a collection of people throughout the workflow and all of us use the same tool. And I think the latter is way…”
Halliday: ChatGPT will select specialized tools over general web scraping for queries
“If Chaggy BT wants to know about congestion in Barcelona or the best restaurant to go in Copenhagen, they are going to start to form an opinion of which tool is the best for the job to answer that, which can give The model, the best quality response and allow …”
Halliday: Brands must audit and participate in offsite platforms influencing ChatGPT
“And I think if you were a brand in that same situation, you would really think about what are the offsite, like not, not on your own domain. What are all the pages that influence the chat GPT for those priority questions as well? So we do a full audit of that.…”
Halliday: Effective AEO requires structuring content to spoon-feed answers directly to LLMs
“There's best practices around structuring that content to make sure you're basically spoon feeding those answers to the models and preparing them for citation. Cause ultimately you want to be quoted a nice big paragraph in a chat GPT response saying on the pro…”
Walker: Startup categories crowd far faster post-ChatGPT than pre-pandemic
“All of these categories are getting so crowded so quickly. And I think that is a real step change from what was going on certainly pre-pandemic, but even just pre-ChatGPT. It just feels very difficult to find a truly novel idea.”
Jain: Narrow AI tasks fail to capture imaginations like open interfaces
“If you go after things that are extremely narrow, then people don't have their imaginations working
On Overdrive, right? I think when Google said we can automate reservations, it didn't spur everyone to go, how do I go apply this everywhere in the world? But …”
Jain: ChatGPT turned workplace search from a vitamin into a painkiller
“And this is where, you know, the transition happened from the vitamin to the painkiller. When people saw that if only I had something like ChatGPT inside my company or all of my company knowledge, people realized that, whoa, like that, no, this is something I …”
Fox: RLHF and dialogue crossed the capability threshold that triggered LLM adoption
“It was the RLHF and the dialogue component of ChatGBT that triggered the takeoff. And that was the capability threshold. Like it was at that time that this capability threshold passed where now it is a prior, like LLMs are a priority for every organization.”
Zima: Early ChatGPT was uninteresting and ineffective for data analytics
“I used early ChatGPT for data, and I was sort of like, this is not that great. It's not very interesting. I'm not going to use this. And so I'd say for six months, I was just completely unconvinced”
Halliday: Sam Altman Convinced Him to Explore LLMs Pre-ChatGPT
“So I was Talking to Sam Altman before ChatGPT was a thing. And I said, like, hey, what's got you excited at the moment? Or some like vague question like that. And he, he's like, the AI stuff's getting really good. And I had no frame of reference for LLMs.”
Soham Ganatra uploaded VC meeting transcripts to ChatGPT to refine his pitch
“I ended up, like practicing it with Chachapiti, which is, like, I don't know, like, something, like at least, like, most of my friends now do. It's, like, I had all my transcripts with all the VCs, In a single job jeopardy session. And I used to say, like, wha…”
Gemini outperforms ChatGPT for some tasks by leaning into UI elements
“I've actually started to find Google's Gemini better than ChatGPT in certain situations because Gemini leans more into a UI, into a Google search.”
Kausas: At least 10 startups in Pylon's YC batch built AI support bots
“So there were at least, like, 10 companies in our batch who were doing just that.”