-->
@Tejretics
I've not done an online debate since DDO. I am potentially interested in the following;
The U.S. should ban fracking.
Debate me!
Posts
Total:
51
-->
@Kasmic
I’d love to debate either of those topics! Which one do you prefer?
-->
@Tejretics
- Artificial intelligence poses a higher risk of causing human extinction than climate change. (Pro)
How likely do you think extinction by AI is? If you think it is likely enough to worry about then I would be willing to take con on this if it was reworded to only be about AI in some way and exclude the comparison to extinction by climate change.
-->
@Discipulus_Didicit
Hmm yeah I would maybe debate that. Admittedly I'd be constrained by my lack of knowledge about ML/computer science, so maybe this would be better suited to a conversation than a debate.
-->
@Discipulus_Didicit
But for what it’s worth, I think it is:
- Pretty unlikely
- Still likely enough to worry, because extinction or a similar catastrophe would be so bad even a small probability of it merits massive concern
-->
@Tejretics
That's reasonable enough. What are your thoughts on AI in regards to the 'technological singularity' idea?
Inevitable/likely/unlikely/impossible?
Soon/far future/impossible to predict?
Will happen overnight/will happen over years/will happen over a generation?
Inevitable/likely/unlikely/impossible?
Soon/far future/impossible to predict?
Will happen overnight/will happen over years/will happen over a generation?
-->
@Discipulus_Didicit
I think people tend to overstate how quickly and how much of an effect AGI will have.
I think rapid economic growth is plausible if an AGI successfully uncovers how to simulate distinct and intelligent humans (in the same way we have kids, albeit on a computer), because then, models of growth that tie exponential growth to population may come true. I don't think an AGI will figure out how to do that especially quickly.
I'm unclear how fast the takeoff from human-level intelligence to transformative AI will be, but I think TAI (i.e., AGI that has the capacity to bring about large-scale social/economic change) seems likely to happen this century (for example, that's roughly the estimate of forecasting TAI using biological anchors). I don't think that means a technological singularity happens this century (and I'm uncertain about whether that's even a coherent concept). I don't think TAI will immediately bring us to a ridiculous futuristic utopia; I think it's more likely to be on the scale of many other significant technological changes over the past two centuries.
However, I think there's a very real risk of TAI killing us all, or locking in bad institutions, or bringing about massive economic inequality. I think it takes much less to do that than to bring about a technological singularity.
-->
@Discipulus_Didicit
I would describe myself as much more of an AGI skeptic than most people concerned about AGI (e.g., Bostrom), but much more concerned about AGI than the median ordinary person.
-->
@Tejretics
I don't think that means a technological singularity happens this century (and I'm uncertain about whether that's even a coherent concept).
It really isn't, which means that boringly enough we seem to mostly agree.
-->
@Discipulus_Didicit
It really isn't, which means that boringly enough we seem to mostly agree.
I still feel a bit uncertain.
In general, it seems like:
- There are limits to economic growth (e.g. fundamental physical limits) in the next ~500 years (until we figure out how to beat those limits)
- Once capital starts to depreciate and diminishing marginal returns of physical capital set in, per capita economic growth is primarily driven by technological change
- Population growth brings a significant share of economic growth driven by technological progress
- A sufficiently advanced AI can “replicate” the effect of population growth, increasing the rate of growth enough to reach the limit much earlier than generally expected
So I find this type of argument plausible. And maybe it’s a steelman of the “technological singularity” folks.
-->
@Tejretics
Specifically I mean the idea that there could be an AI that somehow improves itself or writes the code for a significantly more powerful AI from scratch that then snowballs uncontrollably into exponentially more and more powerful AIs coming into existence without human input being required. The type of things you describe are things that could happen as a result or just happen as a result of natural non-singularity-related advancements in AI.
My claim is that this self-improving AI run away snowball effect idea is silly.
-->
@Discipulus_Didicit
My claim is that this self-improving AI run away snowball effect idea is silly.
And that is what they'll end up winning due to, if enough people take this approach.
-->
@Discipulus_Didicit
My claim is that this self-improving AI run away snowball effect idea is silly.
Have you looked at how Kurzweil's law of accelerating returns has stood up since the beginning of human history
-->
@Incel-chud
Never heard of it.
-->
@RationalMadman
You seem like an intelligent enough guy. Would you be interested in a debate on the practicality of the kind of technological singularity I described?
-->
@Tejretics
I’ll read it and post there later today! But for what it’s worth, in developed countries especially, I think low birthrates are concerning. I think they’re especially concerning for long-run economic growth. More people means more ideas and more technological progress, which is the path to economic growth in countries where simply “catching up” through capital accumulation is no longer possible.I like policy proposals such as child allowances that make it easier to have kids. I also like policy proposals by liberals to lower the cost of living, like building a lot more market-rate housing. I’m uncertain about how to change culture in ways that makes having kids more widely-accepted. I also like a lot more immigration, to temporarily mitigate some of the consequences of low birthrates. It isn’t quite a permanent solution, though, and it doesn’t fix the problem at its root.I think the picture in developing countries is a bit more complicated, because the relevant question is how it affects catch-up growth. India’s TFR looks likely to drop below replacement by 2030, which seems like losing an important “demographic dividend” to engage in labor-intensive manufacturing, which is scary. On the other hand, there’s some research – such as this one, by David Weil and others – suggesting that increases in population density hurt people’s quality of life in developing countries.
I think India is actually below replacement already. Isn't that crazy? The post-2015 crash in births throughout a lot much of the world was incredibly swift. It could reverse, but so far there isn't really a precedent for that.
I agree with almost everything you said in the highlighted bit, those policies do seem to help a bit but overall its a cultural thing. Fascinating to think what the world could look like if trends don't change. Of course they will, but how much and when? Who will inherit the Earth? On immigration what I find really interesting is that because migration is now global (instead of mostly to only a few countries) and with birth rates crashing I think the remaining net emigration countries are going to drain incredibly quickly over the next 20 years. Look at places like Romania or Bulgaria, they are just so hollowed out. A lot of countries like USA or UK or France have decent birth rates and can make up any gaps in the labor market with small (comparatively) amounts of immigration but I don't see how countries like Italy or South Korea can possibly recover when the boomers retire. How is this anything other than death? https://upload.wikimedia.org/wikipedia/commons/thumb/4/4a/South_Korea_Population_Pyramid.svg/1200px-South_Korea_Population_Pyramid.svg.png
-->
@Discipulus_Didicit
The law of accelerating returns is why we get the runaway AI. Kurzweil noticed a type of pattern in history where information technology would double every so often. Moore's law was an expression of this law of accelerating returns and now that we are reaching the end of Moore's law, AI is doubling in capability really fast because of some tricks.
This is specific to information technology and I am struggling to explain the concept, but the law of accelerating returns is why we would have a runaway AI. Technically we could reach a singularity without AI though. Singularity is just one word for the phenomenon. Another might be "intelligence explosion". It's called a singularity, because we can't make accurate predictions beyond that point.
The point is that informations technology increases at a predictable rate. It went from doubling every 10k years in the middle ages to every 100 years as you can see from the year 1900 to 2000. In the year 2000 it was doubling every 6 months.
Once the singularity occurs the doubling will be even faster. Maybe every minute. Kurzweil explains it better. This is my best attempt.
However even if this runaway AI doesn't happen, I still think we are on pace to hit an explosion of intelligence before my 100th birthday and probably somewhere around my retirement.
Here is kurzweil's explanation which will make mine look stupid, but I wanted to make an honest attempt at explaining. https://www.kurzweilai.net/the-law-of-accelerating-returns
-->
@Discipulus_Didicit
No, all you have to do is what Oromagi did vs me in the AI cars debate (and there it was even steeper and more able for him to do it than for you in that debate) just pure defence and cowardice, no need to prove anything. Also I am busy, I only am just about able to handle the Feb tournament as it's 1 debate at a time.
I'd rather do this after mid-March
-->
@thett3
I think India is actually below replacement already. Isn't that crazy? The post-2015 crash in births throughout a lot much of the world was incredibly swift. It could reverse, but so far there isn't really a precedent for that.
Last I checked, TFR in India was like 2.1, and it was predicted to be below replacement by 2030, though maybe something changed since I last checked.
I suspect the crash in births was mostly just a consequence of countries growing? Like the current birthrate is probably the “natural rate” at current income levels, holding constant pre-2015 culture or something.
I agree with almost everything you said in the highlighted bit, those policies do seem to help a bit but overall its a cultural thing. Fascinating to think what the world could look like if trends don't change. Of course they will, but how much and when? Who will inherit the Earth? On immigration what I find really interesting is that because migration is now global (instead of mostly to only a few countries) and with birth rates crashing I think the remaining net emigration countries are going to drain incredibly quickly over the next 20 years. Look at places like Romania or Bulgaria, they are just so hollowed out. A lot of countries like USA or UK or France have decent birth rates and can make up any gaps in the labor market with small (comparatively) amounts of immigration but I don't see how countries like Italy or South Korea can possibly recover when the boomers retire. How is this anything other than death? https://upload.wikimedia.org/wikipedia/commons/thumb/4/4a/South_Korea_Population_Pyramid.svg/1200px-South_Korea_Population_Pyramid.svg.png
Yeah. I’m more worried, actually, about the implications on long-run economic growth than on deficits or the size of the labor force. Chad Jones has a paper on this, in which he models the effect of declining populations on growth. It doesn’t look good.
-->
@Discipulus_Didicit
It seems plausible to me that an AI could self-improve until it becomes much more advanced.
I’m unclear how quickly such a takeoff would happen, though.
-->
@Tejretics
Last I checked, TFR in India was like 2.1, and it was predicted to be below replacement by 2030, though maybe something changed since I last checked.I suspect the crash in births was mostly just a consequence of countries growing? Like the current birthrate is probably the “natural rate” at current income levels, holding constant pre-2015 culture or something.
Paywalled but India fell to about 2.0 last year: https://www.economist.com/asia/2021/12/02/indias-population-will-start-to-shrink-sooner-than-expected However it's possible that it could slightly increase when coronavirus is over, or there could be late registrations that push it up to 2.1 for the year, but the long term trend is clear. I'm assuming you're about 20, when you were born it was 3.3, when I was born in was closer to 3.7. I assume the average emigrant is mid 20s to mid 30s, so were born when women had roughly twice as many babies as they do today. And India is developing rapidly. It seems like the appetite to emigrate will be much much lower really soon. India is such a behemoth that its trends alone are going to carry massive implications for immigration almost everywhere. And this trend is happening throughout the world. Countries with net-emigration are going to drain incredibly quickly imo, and the entire narrative that both sides of the argument have about immigration (basically that there are massive hoards of people champing at the bit to move to developed countries and this is either good or bad) is going to go up in smoke. My prediction anyway, and I'm not very smart. But will be interesting to see. I'm curious to see what you expect to happen
Yeah. I’m more worried, actually, about the implications on long-run economic growth than on deficits or the size of the labor force. Chad Jones has a paper on this, in which he models the effect of declining populations on growth. It doesn’t look good.
Hard to see how it wouldn't totally crush growth. The countries that started having low birth rates early (like Japan, Italy, Spain) haven't exactly put forward very impressive numbers the last few decades. But I don't think we have seen the worst of it yet. In a country like Spain, people in their 40s and 50s outnumber the generation that will be entering the workforce when they retire roughly 2 to 1. How that is going to end in anything other than complete disaster I have no idea. Anyway I bring this up not to rant about my ideas more, but because I think it throws up some extremely serious road blocks to a lot of facets of the current world order like migration, endless economic growth, and old age pensions and I don't know what to expect in the future