February 2025
As an insider working at companies building state-of-the-art AI, I have some non-consensus views about how we should adapt to and co-evolve with them.
First, almost everyone I know, including the smartest, most productive, and wealthiest people are not using AI enough. What is enough? Daily to hourly. At some point, that will be continuous. What does continuous mean? I once had a curmudgeon entrepreneur mentor point out that he built his TV network media company over 25 years and then walked away after putting it on autopilot. He checked in maybe monthly for a few hours, but the company continuously ran to create media services and sent him a large annual dividend (in excess of $100mn in good years). The machine did the work; he benefitted. Today we can use AIs hourly as large reasoning models. When current embryonic agentic systems start to get better, we can embark on continuous usage. I already see a skills gap forming. One interesting labor market study recently suggested that generative AI systems are most used by the wealthiest, best-educated workers.
Second, if you know how to use AI well, you can measurably improve your agency, that is, your ability to get things done with increased intelligence, productivity, and wealth. The catch is that in these early days, no one knows how to use AI well. So that means you need to constantly experiment with doing your daily tasks, chores, and jobs better with AI. Only you can become the expert in defining what “well”, “quality”, or “completion” means. Human taste and judgment matter. Intelligence can be defined in many ways (intellectual, social, artistic, etc), as can wealth (time, social connections, mental clarity, physical health, and finances). AI is a tool that can amplify human capabilities, giving individuals "superagency" – the ability to achieve more with greater intelligence, productivity, and reach.
Third, the upcoming AI age will be the biggest event in the history of homo sapiens as an alien species grows up. I expect it will be bigger than us transitioning to agriculture, forming cities, developing writing or the printing press, creating governments, machines, and industrial factories, developing electricity and electronics, advancing chemistry and biology into engineering and healthcare, or creating the internet and IT world we live. As one example: The printing press increased literacy rates from 30% to 60% in Renaissance Europe, leading to great technical and economic accomplishments (but also war), while some project that AI could automate 40% of cognitive tasks by 2030.
While many want to regulate AI or opine about what governments should do, I believe it will increase the agency and leverage of individuals by a massive amount - that we will have 10x to 100x professionals. We are early, as some recent studies suggest even today’s weak AI can assist coders to complete tasks 55% faster or that consultants using AI produce 40% higher-quality reports and 25% faster. My main goal is to figure out how to increase my own agency and help teach everyone else how to increase theirs. We are preparing for a new species to evolve and spread on our planet - one that is as smart and capable (or more so) than us. Our main goal should be to align it with our noblest and highest values and at the same time learn from it and partner with it to solve our greatest problems (eg poverty, disease, climate change, individual liberty, loneliness and lack of community, etc).
Here are three implementable strategies on how to “merge with AI”, that is, how to partner with it to increase your agency and enrich your life:
1) Use multiple, paid AI systems, and learn the ins and outs of each: Most people I know may use a single LLM like ChatGPT, and then rarely probe it deeply. What I suggest people do is use 3-5 systems and figure out what is the best for each task, chore, or job they’re trying to do. Pay the monthly fee to get the best system - it’s the cheapest subscription you will ever have. My main process is simple - for each task, I test at least 2-3 systems on variants of the task and learn which is the best one at the moment. I then use that one and may repeat the task in 2-3 months with another 3 systems. The ones I use the most and pay for are:
OpenAI GPT-4o or o1/o3: For general answers to questions or discussions, either for fast responses with 4o or some deeper reasoning with o1/o3, or deep research
Deepseek R1 on Perplexity: For searching for information on the web, or for a first cut on searching for products and services (Gemini often comes next)
Gemini 2.0: To interact better with images, books, videos, maps, or audio - like summarizing a podcast or video or talking about it, or deep research
Grok 2.0/3.0: For getting information on real-time news (which is often on X) or searching for variant opinions or creating divergent images.
Midjourney: For general image creation
ElevenLabs: For creating human speech and audio (and hopefully music soon)
EmergentMind: For staying on top of cutting-edge AI research and papers, and asking specific questions to improve my knowledge on how AI works and is evolving
Note that this list will change every 3 months as I experiment with new systems and swap out old ones that fall behind for newer ones.
2) Learn how far the AI can get, how to complement it, and be the quality bar enforcer: For most tasks and jobs, AI can’t do excellent end-to-end work yet, but can maybe get 40-90% of the way there. If you learn how to write detailed and very specific prompts based on what you want, the systems usually do much better. However, the catch today is if you’re an expert in the area, you still need to define and hold a quality bar for the AI systems, which act like junior colleagues or undergraduate student helpers. Current systems aren’t great at agentic planning and doing long sequence tasks, but I expect they will get vastly better every 6 months. You need to plan for that and grow with the improvements. Finally, note that AI can be used in every part of your life that requires intelligence: managing your time, creating a better social life with relationships and community, doing better at work or creating a business, improving your health, understanding the news, trends, and demographic changes around you, etc. Only your imagination and willingness to experiment are the limits.
3) Use AI to help identify and understand your ikigai. Ikigai (生き甲斐) is a Japanese concept that roughly translates to "a reason for being." It's often visualized as a Venn diagram with four overlapping circles:
What you love: Your passions and interests.
What you're good at: Your skills and talents.
What the world needs: Problems you want to solve or contribute to.
What you can be paid for: Work that provides financial stability.
The intersection of all four circles is your ikigai – your purpose in life, the thing that makes you want to get up in the morning. It's not just about finding a job you enjoy, but about finding something that brings you joy, utilizes your skills, contributes to the world, and provides for you. If you find your ikigai, you can focus your life more and push from high-agency into super-agency,
Here are some prompts to get you started:
Life coach: I'm trying to find my ikigai, my reason for being. Could you act as a life coach and guide me through this process? I'd like to explore the four key areas: what I love, what I'm good at, what the world needs, and what I can be paid for. Let's start with what I love. What questions should I ask myself to identify my passions and interests?
Feeling lost: I'm feeling a bit lost and unfulfilled. I've heard about ikigai, the Japanese concept of finding your purpose, and I'd like to explore it. Can you act as a life coach and help me on this journey? I'm not sure where to start. Maybe you could ask me some questions to get me thinking?
Career focus: I'm interested in finding my ikigai. I'm a [Your Profession/Situation - e.g., software engineer, recent graduate, stay-at-home parent]. I enjoy [Mention a few general interests - e.g., working with my hands, creative writing, helping others]. Can you act as a life coach and help me explore how these interests might connect with my skills, what the world needs, and potential career paths?
Values focus: I'm trying to discover my ikigai. I believe my values are important to this. Can you act as a life coach and help me identify my core values? What questions should I ask myself to understand what truly matters to me? How can these values guide me toward finding my purpose?
My final call to action is to take your personal use of AI seriously: Start tomorrow with one AI-augmented hour - use GPT 4o/o3 or Claude 3 to draft emails, Perplexity for research, and R1 for meeting prep. Measure your 30-day productivity gain against a baseline. Embrace AI and try to use it for everything you do that requires intelligence. My main tips:
1) Use multiple, paid AI systems, and learn the ins and outs of each
2) Learn how far the AI can get, how to complement it, and be the quality bar enforcer
3) Use AI to help identify and understand your ikigai.
Evolution favors the augmented.
Arun, you are on an impressive mission to increase your agency and spread AI knowledge. Found myself wanting more of the typical Arun posts with great details on how to!
I am very curious why Claude is not part of your paid tools list? I have recently been using it to dig deeper into technical questions and find it better than Perplexity Pro or Chatgpt Plus.
Also, loved the Ikigai concept, definitely going to try (though not setting high expectations, haha).