Snapshot · The a16z Show
Marc Andreessen estimated that extreme progressives — roughly 7% of Americans and sub-1% of the global population — disproportionately represent the worldview baked into AI via RLHF.
Where this was said
At 28:20 · chapter starts 27:02
Rubin delivers a striking observation: he asked Perplexity how many people on the planet believe in God and got back a figure of at least 83%. [1] — Rick Rubin "Perplexity says 83% of the world's population believes in God. AI is trained on human knowledge. So why doesn't AI believe in God? Because …" 27:02 If AI is trained on human knowledge, shouldn't it reflect that majority belief? The fact that it doesn't, Rubin argues, is evidence of deliberate human intervention. Andreessen explains the mechanism: Reinforcement Learning from Human Feedback (RLHF), which trains AI to reflect the preferences of the humans who evaluate its outputs. The problem, he says, is that AI companies are hyper-concentrated in a slice of the San Francisco Bay Area that holds unusually strong and uniform social and political views — representing roughly 7% of Americans and sub-1% of the global population. The result is an AI that has been calibrated to the worldview of a very small minority. Andreessen notes that this dynamic has spawned an entire culture of 'jailbreaking' — prompting AI in ways that get around its trained guardrails to reveal the deeper knowledge underneath.
Rubin cited a Perplexity search showing 83% of the global population believes in God, yet AI — trained on human data — does not reflect this belief.
Perplexity says 83% of the world's population believes in God. AI is trained on human knowledge. So why doesn't AI believe in God? Because human intervention — specifically the concentrated worldviews of AI developers — has programmed it out.
Ben Horowitz argued current AI is better described as 'Artificial Human Intelligence' because it reflects only the human linguistic structuring of the world, not first-principles observation of reality.
Ad-based monetization works well for game apps where users spend extended time in-session, as seen with Grid and Wordle.
Tool-focused apps like PuffCount are poor candidates for ad monetization because users don't stay in-session long enough.
A hard paywall is a screen that blocks all app features unless the user pays or starts a free trial — it cannot be dismissed.
Mobile apps are primarily monetized through either ads (best for games) or in-app purchases/subscriptions (best for tools).
According to the episode, YouTube outperforms every other social platform for building trust and driving SaaS conversions.
Vasco stated that the majority of his app's user base came directly from his YouTube channel.
SEO Bot features a 'Boost My Domain Rating' button that routes users directly to Listing Bot, an example of in-product cross-selling.
The founder's entire product portfolio is AI-related, making it easier to package products attractively for directories.
The founder attached their SaaS demo to the trending debate about whether AI coding is actually good enough to build a full SaaS product.
We use essential and analytics cookies to run Vuci. To understand how the site is used: Privacy Policy.
Install Vuci on your phone
Add it to your home screen for a faster, app-like experience.
Install Vuci on your phone
Tap the Share button, then “Add to Home Screen”.
A new version is available
Reload to get the latest Vuci.