GPT-4
A large language model from OpenAI representing a major capability jump, used in products like Microsoft Bing search.
As defined in Zvi Mowshowitz on AI and the Dial of Progress
Where it comes up · 22
- Why Vertical LLM Agents Are The New $1 Billion SaaS Opportunities7 claims
- AI & GPT-4 Revolutionize Education With Sal Khan | EP #35 Moonshots and Mindsets6 claims
- Bill Gates Reveals Superhuman AI Prediction4 claims
- Geoffrey Hinton - Two Paths to Intelligence4 claims
- Marc Andreessen on Why AI Will Save the World3 claims
- Zvi Mowshowitz on AI and the Dial of Progress3 claims
- OpenAI’s warning: Washington isn’t ready for what’s coming2 claims
- Conversations with Tyler 2023 Retrospective2 claims
and 14 more
Most important claims · 10 of 52Search all claims for “GPT-4” →
- When Sal Khan first saw GPT-4 in early August 2022 (three months before ChatGPT's public launch), he recognized it was substantially more capable than GPT-3 and tested versions, prompting him to email OpenAI leadership asking why they were launching so early, though the leadership clarified they hadn't launched the model itself—only ChatGPT, a chat interface on top of it.AI & GPT-4 Revolutionize Education With Sal Khan | EP #35 Moonshots and Mindsets
- In high school, Gates was intrigued by AI capabilities like Shakey the robot at Stanford Research Institute, which could engage in reasoning and create execution plans; he initially believed speech recognition and image recognition would be fairly solvable problems, but the Holy Grail was the ability to read and represent knowledge like humans did—something nothing was good at until GPT-4.Bill Gates Reveals Superhuman AI Prediction
- OpenAI partnered with Khan Academy for two strategic reasons: (1) to benchmark GPT-4's knowledge on standardized tests like the SAT and AP exams where Khan Academy has tens of thousands of test items, and (2) to launch GPT-4 with a handful of organizations that could demonstrate positive social use cases.AI & GPT-4 Revolutionize Education With Sal Khan | EP #35 Moonshots and Mindsets
- When Jake Heiler first encountered GPT-4 under NDA from OpenAI, he made the decision within 48 hours to pivot the entire 120-person company to build Co-Counsel, shifting everyone from their existing product roadmapWhy Vertical LLM Agents Are The New $1 Billion SaaS Opportunities
- Before ChatGPT's public release, CaseText had early customer testing of GPT-4 under extended NDA with OpenAI, allowing law firms to experience the AI for the first time without knowing it was GPT-4, and they observed tasks that normally took a full day taking only 90 secondsWhy Vertical LLM Agents Are The New $1 Billion SaaS Opportunities
- CaseText went from $0 to $100 million valuation in the first 10 years, then to a $650 million exit to Thomson Reuters just two months after launching Co-Counsel on GPT-4Why Vertical LLM Agents Are The New $1 Billion SaaS Opportunities
- Embedding a book inside a large language model like GPT-4 is the future of writing and reading: it lets readers summarize chapters, extract anecdotes, or ask how a scholar might disagree, and over time the AI component will become so much stronger that most things one buys or reads, including op-eds, will have queryable embedded versions available at the margin for free.Tyler Cowen on the GOAT of Economics
- On the HumanEval coding benchmark, GPT-3.5 with zero-shot prompting achieves 48% accuracy, GPT-4 achieves 67% accuracy, but GPT-3.5 wrapped in an agentic workflow actually outperforms GPT-4, and an agentic workflow around GPT-4 performs even better.What's next for AI agentic workflows ft. Andrew Ng of AI Fund
- OpenAI runs standardized exams (SAT, ACT, and others no longer permitted for college admissions) through each new GPT version, and because these exams are essentially IQ tests, this is how one can approximate the IQ of these systems—China replicated this by running its own state exams, which include Marxism and Xi Jinping-thought, through its LLM.Marc Andreessen on Why AI Will Save the World
- GPT-4 was trained on massive internet data including World Bank reports, HBR articles, newspaper articles from Kenya, and other global sources. This training data means the model has surprisingly good knowledge of emerging market contexts like Kenya, knowing terms like 'matatu terminus' (bus terminus) and being able to answer questions about Kenyan dairy prices, though with some data drift since training.The Disruptive Potential of GenAI for Emerging Markets | Rem Koning | Leading with AI