Location leaks into incognitio.
I asked it to tell me what it has in its context and it told me:
>Your approximate location: Bengaluru, Karnataka, IN — this comes from a single line in my system prompt (“User’s approximate location: Bengaluru, Karnataka, IN”), not from anything you told me.
>Rather, it can produce goodness and empirically in the case of human evolution it has created things that we would call goodness.
Hasn’t it produced cooperation and goodness amongst humans who’re of similar power? My intuition pump for AI risk is what humans have done to other beings sharing the planet and the story there is not encouraging.
I agree. Limited resources drives the need to have abstractions. Communication probably simply provides another drive to abstract at even higher levels.
I would imagine for many animals evolution would have pre-programmed abstractions like “object” or “fruits” so they come pre-parsed.
Building devices is actually like setting up physical experiments. If they do what they’re supposed to do, you can increase your confidence in the mechanisms that explain how they work.
Many of the neuroscience discoveries happened via a mix of luck and detective work—which is what I took from the post as a way for progressing science of complex systems.
Yes, you’ve paraphrased the argument well.
While progress for any certain technology can certainly be exponential, the benefit or impact of it may not be exponential if we assume that the lowest hanging fruits (in terms of cost/benefit tradeoff) gets plucked first, and then it becomes harder and harder to derive economically relevant benefit even with tech progress.
What unbounded benefit relies on whether we can keep on discovering new types of general-purpose technologies, and not just an improvement in current technologies.
E.g. faster air travel depends on some new uncertain technology to be invented. We can only squeeze incremental benefits from further improving the current air travel technologies.
Also see Eroom’s law which suggests declining R&D productivity for drug discovery.
Whether we can keep on discovering such general-purpose technologies is pure speculation, so there’s some probability that we may actually not be in a weird cross-roads within the next 200 years.
This is interesting. I agree that the various cleanup technologies add to the economic growth.
My point was more suggestive in the sense of whether that poses a practical constraint to economic growth (in terms of say negative feedback loops) when the cost of cleanup outstrips the benefit derived (and hence putting an upper limit to the desire of the benefit).
Source?
Truth is multi-faceted notion, when it comes to its relationship with science. Going with the standard structural realism arguments, there’s a) ontology (entities that exist); b) relationships / structure (between entities).
The ontology in our theories is regularly usurped by newer theories (we don’t believe in ether or infinite space time of Newton anymore so we should expect ontology of current theories to not hold true as well). But something does get preserved and it’s the relationship between coarse grained entities (F=ma will more or less hold true even if our notion for what mass is changes over time).
The implication of this distinction in any epistemology is that the “true” reality will not be — could not be —revealed to us. We only have observations to rely on. What we can expect to model is observed regularities. What models get adopted is sociological phenomena to a great extent but the scientific community weighs risky, precise predictions that turn out to be true a lot more because it’s an extremely strong evidence of the model having captured some aspect/view of true regularities of the underlying reality. (But, of course, future prediction is not the only criteria—models get adopted for all sorts of reasons including unification of disparate observations).
Overall, I’m unsure if the notion of truth is needed at all. What we need is useful models and interpretable, extrapolative models are more useful than black box models.