“it would send it off to GPT3 at the time.”
GPT-3
By OpenAI
1 recommend/use · 7 sourced episodes
Every sourced reference
Short attributed excerpts only. Timestamps are approximate.
“And then GB3 came out in 2020”
“Prior to Anthropic, he was one of the architects of GPT3 at Open AI.”
“this is GFD3 you know instruct doesn't exist”
“until like I saw gpt3”
“we've played around with gpt3 so gpt3 is basically generating sentences that are similar to other sentences”
“Codex which is a derivative of another kind of one that you might be familiar with GPT-3.”
Related frameworks
Adaptive Evaluation Over Static Benchmarks
Measure AI robustness with attackers that learn, not with a frozen dataset of yesterday's attacks
Ambitious Retry (Pass@N) Tool Use
Get more from AI tools by asking for the ambitious change and fully restarting on failure instead of hammering the same attempt
Build for the Exponential (Skate to Where the Puck Is Going)
Build products for the model that arrives in 6-12 months, betting that today's 20%-working features hit 100%
Build the Company Only You Could Build
Reject the pivot-and-blitzscale playbook; commit to the one idea that dies without you.
CaMeL Permission Pre-Restriction
Grant an agent only the permissions its stated task needs, decided before it runs
Chop-It-Up AI Collaboration Loop
Don't hand AI one giant spec — specify a little, review a little, repeat in tight loops.
Complement-the-Frontier Model Strategy
Don't rebuild foundation models — train small specialty models that attack their weaknesses in speed, cost, and niche tasks.
Constitutional AI Self-Critique Loop
Align a model to written principles by having it critique and rewrite its own outputs, then train on the fix
Deep Human Evaluation Over Leaderboards
Measure real progress with experts who work through the answer, not crowds who vibe for two seconds.
Design For Tomorrow's Majority, Not The Vocal Minority
Optimize for the far larger future user base, then manage the unhappy few with authentic listening.
Design the Dream Objective Function
You become what you measure — so measure the rich thing you actually want, not the easy proxy.
Differentiate Above The Model, Not Around Its Gaps
Build your moat outside the LLM so your product gets better as the models get better.
Dogfood-Driven Realism
Be the end user, use your product intensely daily, and never ship anything that isn't useful to you.
Earn The Next Role By Driving Impact In This One
Stop eyeing the next promotion; be judged on impact delivered in the job you already hold.
Editorial SEO Operating Model
Split the work: SEO manager picks topics and outlines, a domain-expert writer writes, then analyze.
Expected-Conversions Prioritization
Rank topics by search volume x topical authority x conversion intent, not by volume alone.
Fall-On-Your-Face Taste Calibration
Deliberately push AI past its limits in a safe environment to build a gut feel for what it can do.
Flow-Preserving AI Assistance Design
Design AI suggestions around the user's flow state: no panel switches, no waiting, ephemeral by default
Follow People, Not Plans
Choose jobs by who you'll learn from, not by a five-year plan or a financial bet.
High-Ceiling Market Test
Attack markets with a high ceiling and easy switching; incumbents win only where there's little left to build.
Hypothesis-Driven SEO
Generate hypotheses by analyzing sites (and asking their SEO managers), then test to validate.
Internal Link Spread Optimization
Most sites starve their own pages of internal links; fixing spread is SEO's biggest quick win.
Knowledge-Work End-State Idea Generation
Pick an area of knowledge work and reason forward to its end state as AI matures — then build for that.
Marginal-Impact Reasoning Under Tail Risk
Prioritize catastrophic low-probability risks by expected value and by how few people are already working on them
Persona-Framing for AI Product Behavior
Pick a human metaphor for your AI, then derive its behavioral guardrails from that role
Reinforcement Learning Environment Design
Build a fully-fleshed simulated world, inject real chaos, and reward the trajectory, not just the answer.
Research-to-Product Graduation
Move an incubated moonshot from the research lab to a product team without killing it or trapping the researchers
Resting in Motion
Treat the busy, working state as your normal baseline so heavy long-term work stays sustainable
SEO Readiness Assessment
Two questions decide if SEO is worth investing in: is the market big, and do you have authority?
SERP Reverse-Engineering
To learn what page to build, read what Google already ranks for your target keywords.
The 60/30/10 Portfolio Capacity Split
Allocate team capacity across incremental wins, operations, and audacious bets before priorities compete
The AI Deployment Risk Triage
Classify any AI deployment into one of three risk tiers before spending a dollar on defense
The AI Security Vendor Due-Diligence Test
Five questions that expose whether an AI guardrail vendor is selling real protection or theater
The Angry-God Containment Lens
Assume the AI is a malicious agent trying to hurt you, then engineer so it structurally can't
The Complementary Learning Modes Ladder
Mastery needs many kinds of practice — imitate a master, get preference feedback, get graded, then live it.
The Economic Turing Test
Measure transformative AI by whether you'd hire an agent for a real job without knowing it's a machine
The PM-Engineering Alignment Operating System
Run product and engineering as one leadership unit with clear ownership and async, iterated reviews.
The Teach-Me-Something-In-One-Minute Interview
A 60-second timed teaching exercise scored on completeness, complexity, and clarity
The Three Types of SEO
Programmatic, editorial and technical SEO are different games; pick the one your queries reward.
The Two-Sided Quality Signal
Great curation does two opposite jobs at once: kill the worst, surface the best.
Three Worlds Theory of Change
Decide what to actually do about a hard problem by naming the pessimistic, optimistic, and pivotal worlds you might be in
Throwaway Prototyping To Discover The Product
Build disposable prototypes on real data to feel what works before writing any production code.
Topical Authority Adjacency
Rank by starting where you're already known, then expand into semantically adjacent topics.
Topics Over Keywords
Modern Google ranks one page for a cluster of 200-2,000 keywords; target the topic, not the keyword.
Two-Day Work-Test Hire
Bring finalists on-site for two days to do a real end-to-end project — it screens skill and fit at once.
Use AI As Leverage On Your PM Time
Assign the model a role, feed it more context than you could read, and iterate the prompt until it works.
People in these episodes
Related resources
- 80,000 Hours
- AG1
- Ahrefs
- AI 2027
- Airbnb
- Alexa
- Amazon
- Amazon S3
- Amazon Web Services
- Amplitude
- Android
- Anthropic
- Apple Podcasts
- Apple Terms of Service
- Area 120
- Arrival
- Atlas
- AWS
- Ben's Bites
- Beso Chocolate
- BetterUp
- BigQuery
- Bloomberg Terminal
- Brief Interviews with Hideous Men
- ChatGPT
- ChatPRD
- Chrome
- Chromium
- Claude
- Claude Artifacts
- Claude Code
- Claude Sonnet
- Clearscope
- Coda
- Codex
- Comet
- Contact
- Contentful
- Cursor
- DALL-E
- Datadog
- DeepMind
- Discord
- Don't Write That Jailbreak Paper
- Dotdash
- Dreamforce
- eBay
- Eppo
- Figma
- Fin
- Finland
- Gemini
- Gemini 3
- GitHub
- GitHub Actions
- GitHub Advanced Security
- GitHub Copilot
- GitHub Discussions
- GitHub Next
- Glassdoor
- Glean
- Gmail
- Gödel, Escher, Bach
- Gong
- Good Strategy/Bad Strategy
- Google Search
- GPT-2
- GPT-4
- GPT-5
- Granola
- Graphite
- Grok
- Hack Prompt and Learn Prompting AI Security Course
- Hacker News
- Healthy Sleep Habits, Happy Child
- How to Talk So Kids Will Listen & Listen So Kids Will Talk
- HubSpot
- Hugging Face
- Intercom
- Investopedia
- Kurzgesagt – In a Nutshell
- LaunchDarkly
- Le Ton beau de Marot
- Lemon.io
- Lenny's Newsletter
- Lenny's Podcast
- lennyspodcast.com
- Linear
- Listen
- Llama
- LM Arena
- LucidLink
- Machine Intelligence Research Institute
- Magpie Murders
- Make It So
- MasterClass
- Maven
- Meta
- Metronome
- Microsoft
- Microsoft Azure
- Microsoft Windows
- MIT
- Mixpanel
- Model Context Protocol
- Modern Treasury
- Montessori education
- MySpace
- NerdWallet
- Netflix
- No Priors
- Notion
- OneSchema
- OneSchema FileFeeds 2.0
- OpenAI
- Our World in Data
- Pantheon
- PayPal
- Perplexity
- Product-Led SEO
- Ramp
- Repello AI
- Replacing Guilt
- Robinhood
- Salesforce
- Sauce
- Scale AI
- Screaming Frog SEO Spider
- Search Off the Record
- Semrush
- ServiceNow
- Sidebar
- Similarweb
- Slack
- Slack Clips
- Slack Huddles
- Snowflake
- Snyk
- Sora
- Spotify
- Sprig
- Stable Diffusion
- Stack Overflow
- Star Trek
- Story of Your Life
- Stripe
- Substack
- Superhuman
- Superintelligence
- Surge AI
- Surge AI Blog
- Switch: How to Change Things When Change Is Hard
- Team of Rivals
- Ted Lasso
- The Alignment Problem
- The Cognitive Revolution
- The Everything Store
- The Last Question
- The Memory Palace
- The Myth of Sisyphus
- The Neuron
- Thumbtack
- Ticketmaster
- Travelers
- Tripadvisor
- United Nations Universal Declaration of Human Rights
- Upwork
- Vanta
- Visual Studio Code
- Watershed
- Waymo
- Webflow
- Working Backwards
- YouTube
- Zillow
Spot an error or want this page removed? Request a correction or removal.