3 ms·

In principle we do not know the long-term impact in advance, but we can often make a probabilistic estimate. Choosing career as an AI researcher vs a Go player,

In principle we do not know the long-term impact in advance, but we can often make a probabilistic estimate. Choosing career as an AI researcher vs a Go player, for example. * Note: If you want to make an impact, whichever your career inclination is, this website provides some great guides: https://80000hours.org/ https://80000hours.org/ I think trying to capture more of the world's resources for Google, Facebook, etc. do have a large impact in aggregation, so it's not a trivial pursuit. DeepMind itself and many other techs to come are being funded by these companies. Personal impact may not be extremely large since there are many other brilliant people working on the same thing but some with a leading role could make a contribution there. (For example if you work for Google, which I believe is a major force for long-term innovation.) What I had in mind was people who spend careers on games, puzzles, or gambling. I once watched a Ke Jie interview after his loss to AlphaGo and could tell that he is brilliant and would likely be successful in other careers that require sharp thinking. Some like Christopher Lagan needed to live a double life and is still unlikely to make any real impact on the world. https://en.wikipedia.org/wiki/Christopher_Langan https://en.wikipedia.org/wiki/Christopher_Langan

panic 9y ago
In principle we do not know the long-term impact in advance, but we can often make a probabilistic estimate. I'm not sure this is true. We can't make a probabilistic estimate without a model. And to design an accurate model, you need to describe precisely what it means to ask, "is this activity worth pursuing?" Is there a way to do this without involving all of human society in your model?
- nopinsight 9y ago
  As a first approximation, a simple model could be: The number of people impacted * How greatly their lives are impacted given your existence
659087 9y ago
> I think trying to capture more of the world's resources for Google, Facebook, etc. do have a large impact in aggregation, Capturing more of the world's resources for a small number of massive advertising corporations is not beneficial to society in any way. The "innovation" happening at those companies is mostly detrimental, and focused on finding new ways to surveil and manipulate people for profit.

panic 9y ago

In principle we do not know the long-term impact in advance, but we can often make a probabilistic estimate. I'm not sure this is true. We can't make a probabilistic estimate without a model. And to design an accurate model, you need to describe precisely what it means to ask, "is this activity worth pursuing?" Is there a way to do this without involving all of human society in your model?

nopinsight 9y ago
As a first approximation, a simple model could be: The number of people impacted * How greatly their lives are impacted given your existence

659087 9y ago

> I think trying to capture more of the world's resources for Google, Facebook, etc. do have a large impact in aggregation, Capturing more of the world's resources for a small number of massive advertising corporations is not beneficial to society in any way. The "innovation" happening at those companies is mostly detrimental, and focused on finding new ways to surveil and manipulate people for profit.