January 6th is the day I finished reading Superintelligence, and that’s the day my life felt like it entered protagonist mode. I tore through ten books over the next week and a half, from The Art of Strategy to Thinking: Fast and Slow. I proceeded to finish the considerable remainder of the Sequences I’d left unread. It wasn’t fear that fueled me – it was a sweet blend of curiosity and protectiveness, about and for this cruel and beautiful world in which we live. It was a sense of the possibilities afforded to those lucky enough to live in what Scott Alexander dubbed the carefree springtime of the universe. It was a rekindling of my youthful inquisitiveness, when I’d spend afternoons lost in math books, coding a C++ text RPG, or caring for a caterpillar over its evolution, just to set it free.
Month 2
I applied for the upcoming CFAR workshop and was later accepted (the interview was the first time where I heard another person say things like “updating”). I continued my non-technical reading, summarized a technical alignment result, and reviewed Naïve Set Theory. The concepts in that small set theory book expanded my mathematical horizons beyond measure (okay, I haven’t studied measure theory yet, so not quite beyond measure). This is also when things really clicked for me – I started generating novel insights regularly. Finally, I decided to try for 5 minutes before giving up on my CHAI application; this led to an idea of which I am truly proud.
I do not yet know whether I will be accepted for CHAI’s internship, but the idea merits two posts of its own: one for the proposal itself, and one for the wonderful emotional and psychological process of discovery I experienced.
Month 3
I find myself eagerly reading chapters and completing problem sets from AI: A Modern Approach, well beyond what is required for my class. I’m writing more often, learning more deeply, and generally my hedonic index is through the roof relative to last fall. I know I’ve given the impression of having let up over the last month, but I really haven’t – there are multiple projects of mine whose fruits I look forward to sharing.
Forwards
I’ve noticed that I go through phases of intense interest in activities, commitments, and games – this shift tended to occur multiple times per year, but has slowed as I’ve matured. I don’t expect that to happen here. I sure hope it doesn’t – I’m not ready to relinquish this wonderful lightness, this eagerness to explore how the world really works on a gears level, this internal sense of purpose.
This lightness does not appear to be conditional on future acceptance or recognition – even when imagining a world in which I know with certainty that no AI safety research group brings me into their fold, the light and curiosity remain.
I am so very grateful that I found this community.
Decision
I remember the moment that was, in hindsight, a litmus test of my readiness. I was finishing the Sequences, and the reality of the “dark world” pressed down on me. My chest felt tight and pressed down upon, and gravity seemed so much stronger than usual; the problems were large, and I felt small. Would this become an obsession that would consume my psyche and my studies? Where did I leave my hero license? Who was I to make this kind of status claim, to believe that I could make progress on problems of such importance?
I came to Beyond the Reach of God and I knew I had to decide. Someone close to me had advised me to step back, to remain in the shallows for a while longer – where I was tall enough to stand. I, however, was privy to information which they were not.
Trout can swim.
Gnawing Shadows
I can sense a mix of reasonable dissatisfaction with my performance, and psychologically unrealistic expectations. I’ve taken far longer than I wished on my AI book; if only I were less vulnerable to pica, if I studied an extra hour each day, if the concepts had come to me more easily… I imagine worlds in which I did better and had also made substantial progress in, say, topology or linear algebra by this point in time.
I’ve come so far, but I also am quite aware of the countless levels above mine. Sometimes it weighs on me, but usually I view the task before me with twisted pleasure.
Lightness and Unease
Light
Month 1
January 6th is the day I finished reading Superintelligence, and that’s the day my life felt like it entered protagonist mode. I tore through ten books over the next week and a half, from The Art of Strategy to Thinking: Fast and Slow. I proceeded to finish the considerable remainder of the Sequences I’d left unread. It wasn’t fear that fueled me – it was a sweet blend of curiosity and protectiveness, about and for this cruel and beautiful world in which we live. It was a sense of the possibilities afforded to those lucky enough to live in what Scott Alexander dubbed the carefree springtime of the universe. It was a rekindling of my youthful inquisitiveness, when I’d spend afternoons lost in math books, coding a C++ text RPG, or caring for a caterpillar over its evolution, just to set it free.
Month 2
I applied for the upcoming CFAR workshop and was later accepted (the interview was the first time where I heard another person say things like “updating”). I continued my non-technical reading, summarized a technical alignment result, and reviewed Naïve Set Theory. The concepts in that small set theory book expanded my mathematical horizons beyond measure (okay, I haven’t studied measure theory yet, so not quite beyond measure). This is also when things really clicked for me – I started generating novel insights regularly. Finally, I decided to try for 5 minutes before giving up on my CHAI application; this led to an idea of which I am truly proud.
I do not yet know whether I will be accepted for CHAI’s internship, but the idea merits two posts of its own: one for the proposal itself, and one for the wonderful emotional and psychological process of discovery I experienced.
Month 3
I find myself eagerly reading chapters and completing problem sets from AI: A Modern Approach, well beyond what is required for my class. I’m writing more often, learning more deeply, and generally my hedonic index is through the roof relative to last fall. I know I’ve given the impression of having let up over the last month, but I really haven’t – there are multiple projects of mine whose fruits I look forward to sharing.
Forwards
I’ve noticed that I go through phases of intense interest in activities, commitments, and games – this shift tended to occur multiple times per year, but has slowed as I’ve matured. I don’t expect that to happen here. I sure hope it doesn’t – I’m not ready to relinquish this wonderful lightness, this eagerness to explore how the world really works on a gears level, this internal sense of purpose.
This lightness does not appear to be conditional on future acceptance or recognition – even when imagining a world in which I know with certainty that no AI safety research group brings me into their fold, the light and curiosity remain.
I am so very grateful that I found this community.
Decision
I remember the moment that was, in hindsight, a litmus test of my readiness. I was finishing the Sequences, and the reality of the “dark world” pressed down on me. My chest felt tight and pressed down upon, and gravity seemed so much stronger than usual; the problems were large, and I felt small. Would this become an obsession that would consume my psyche and my studies? Where did I leave my hero license? Who was I to make this kind of status claim, to believe that I could make progress on problems of such importance?
I came to Beyond the Reach of God and I knew I had to decide. Someone close to me had advised me to step back, to remain in the shallows for a while longer – where I was tall enough to stand. I, however, was privy to information which they were not.
Trout can swim.
Gnawing Shadows
I can sense a mix of reasonable dissatisfaction with my performance, and psychologically unrealistic expectations. I’ve taken far longer than I wished on my AI book; if only I were less vulnerable to pica, if I studied an extra hour each day, if the concepts had come to me more easily… I imagine worlds in which I did better and had also made substantial progress in, say, topology or linear algebra by this point in time.
I’ve come so far, but I also am quite aware of the countless levels above mine. Sometimes it weighs on me, but usually I view the task before me with twisted pleasure.