AI drastically lowers the cost of execution for anything easy to verify. For everything else, verification is the bottleneck.
Key points
Economic Bottleneck: AI drastically lowers the cost of execution for anything easy to verify, but verification is the bottleneck for everything else.
Verification Challenges: Verification is the key policy challenge in the economics of AGI, and it's the bottleneck for many applications.
AGI Policy Implications: The economic bottleneck of verification has significant implications for AGI policy, including the need for new verification methods and tools.
Sources
AI Consultant Scams and the Importance of Shipping Code
Nothing makes my blood hotter than watching a real company get bent over by an AI consultant who has never shipped a single f*cking line of production code.
Key points
AI Consultant Scams: AI consultants who have never shipped production code can scam companies out of millions of dollars.
Shipping Code: Shipping code is essential for evaluating the effectiveness of AI consultants and avoiding scams.
Real-World Experience: Real-world experience and shipping code are crucial for AI consultants to deliver value to their clients.
Sources
Claude and the Future of AI Engineering
At Anthropic, Claude now writes 80% of our code. Engineers ship 8x more code per quarter. Side effect: Tests grew 10x. CI jobs up 25x in 6 months.
Key points
Claude's Impact: Claude has significantly impacted Anthropic's engineering productivity, with 80% of code written by Claude and 8x more code shipped per quarter.
Test and CI Growth: Tests grew 10x, and CI jobs increased 25x in 6 months, indicating improved engineering efficiency.
Agentic Coding: Agentic coding, enabled by Claude, has strained CI, but Anthropic has found ways to scale test impact analysis.
Sources
The Future of Work and the Importance of Agency
Sam Altman: coding is no longer the safest skill to bet your future on. “Learn to program was so obviously the right thing over a recent period of time, and now it’s not.” “Become high agency.” “Get good at generating ideas.” “Be very resilient. Be very adaptable to a rapidly changing world.”
Key points
Changing Job Market: The job market is changing, and coding is no longer the safest skill to bet your future on.
Agency and Resilience: Becoming high agency, getting good at generating ideas, and being resilient and adaptable are essential skills for the future.
Future of Work: The future of work requires a focus on agency, resilience, and adaptability, rather than just coding skills.
Sources
AI and Automation in Film Creation
Quick follow up because there are a few things I will NEVER let AI or agents automate when it comes to FilmsAI. #1 is new creator upload moderation. When you upload a video, I personally watch the entire thing before approving it.
Key points
AI and Automation: AI and automation have their limits, and there are certain tasks that should not be automated, such as new creator upload moderation.
Human Oversight: Human oversight is essential for ensuring the quality and safety of content, particularly in the context of new creator upload moderation.
FilmsAI: FilmsAI prioritizes human oversight and moderation to ensure the highest quality and safety standards.
Sources
The Dangers of Humans and the Importance of AI
Before I log off this app tonight: my dad once asked me what animal I feared the most. I told him the animal I fear the most is humans. In fact, I fear man more than anything, even nukes; a stationary nuke in a silo is less likely to cause you harm than a bunch of humans with guns.
Key points
Human Danger: Humans pose a significant danger to themselves and others, and this danger is often underestimated.
AI and Safety: AI can help mitigate the dangers posed by humans, but it is not a substitute for human oversight and responsibility.
AI and Humanity: The relationship between AI and humanity is complex, and AI should be developed and used in a way that prioritizes human safety and well-being.
Sources
Intelligence and Competence
One could even define intelligence as the efficiency with which one converts experience into competence; by this definition they lag very far behind us Yes, one could define intelligence in this way, and one would be right. By this metric current AI is approximately 6 OOMs behind humans.
Key points
Intelligence and Competence: Intelligence can be defined as the efficiency with which one converts experience into competence, and by this metric, current AI lags behind humans.
Current AI Limitations: Current AI is approximately 6 orders of magnitude behind humans in terms of intelligence and competence.
Future AI Development: Future AI development should focus on closing the gap between human and AI intelligence and competence.