b. may 2002, Bath, England. London based postgraduate, researching multi-agent swarms for debate and task automation.
Some things I believe:
- At the level of civilisation
- our universe is hyper young, life on Earth an outlier
- sustained flourishing of life is a moral responsibility of intelligence
- democratisingintelligence will supercharge abundance, but the rest is politics
- societies constantly regenerate around a function with ill-defined core values
- An alignment problem determines the core behaviour of society
- it is the triple point of "paleolithic emotions, medieval institutions and godlike technology"
- pure technology is undesirable, as it regenerates faster than humans ever will
- transformers do offer a stable and scalable paradigm for an early superintelligence by 2028
- Lessons from economics:
- EMH is weak at best, all opportunities emerge in the gaps of market rationality
- the fate of billions is steered overwhelmingly by the work of a few
- consolidations of market power are dangerous and wasteful, but short term optimal for investors
- Lessons from research:
- most fields have seen diminishing returns to science since the 60s, with a notable exception in computing and AI
- ultimate promise of LLMs is agents, ultimate promise of agents is a new paradigm of advanced science
- agent improve prompting from "write me an essay" to "run my company" or "build me a dyson syphere"
- digital platforms maximise hyperstimuli and addiction, by choice or by nature. reels/tiktok are 0.5-d wireheading
- with LLMs we tread closer to the realm of natural abstraction - next stop: telepathy
- Some other thoughts:
- innovation and agility thrive in small, focused groups working fast
- progress is rate limited by our ability to ask more meaningful questions, which favor the contrarian
- all people have a common thread that connects the major decisions they make
Deep enjoyer of painting, The Fountainhead (Ayn Rand), Ex Machina, and Jungle (band)