TE
TechEcho
Home24h TopNewestBestAskShowJobs
GitHubTwitter
Home

TechEcho

A tech news platform built with Next.js, providing global tech news and discussions.

GitHubTwitter

Home

HomeNewestBestAskShowJobs

Resources

HackerNews APIOriginal HackerNewsNext.js

© 2025 TechEcho. All rights reserved.

Complexity No Bar to AI

120 pointsby tmfiover 4 years ago

16 comments

Animatsover 4 years ago
Well, it&#x27;s better than the argument that machines can&#x27;t resolve undecidable questions but humans can.<p>There&#x27;s a large family of problems that are NP-hard in the worst case, but much easier in the average case. Linear programming and the traveling salesman problem are like that.<p>The research question I would pose is, why is robotic manipulation in unstructured spaces so hard? Machine learning has not helped much there. Yet it&#x27;s a fundamental animal skill. We&#x27;re missing something that leads to success in that area. Whatever that is, it may be the next thing after machine learning via neural nets.<p>Note that it&#x27;s not a human-level problem. Primates have the hardware for that. Mammals down to the squirrel level can manipulate objects. Mice, maybe. Mouse-level neural net hardware exists. It&#x27;s not even that big. The University of Manchester&#x27;s neural net machine supposedly has mouse-level power, in six racks.<p>I tried some ideas in this area in the 1980s and 1990s, without much success. Time for the next generation to look at this. More tools, more compute power, and more money are available.
评论 #26217958 未加载
评论 #26218346 未加载
评论 #26217801 未加载
评论 #26218450 未加载
评论 #26227588 未加载
评论 #26219043 未加载
评论 #26218310 未加载
评论 #26231161 未加载
idlewordsover 4 years ago
It&#x27;s past time to start calling these treatises on hyperintelligence what they are—theology—and treating them with the respect they deserve, which is a lot less than they currently get on this site.<p>People have been theorizing about the attributes of the Absolute since forever. Just because you start talking about building a god, rather than positing one already in existence, doesn&#x27;t make the discussions about the nature of such hypothetical superbeings any more fruitful.
评论 #26218337 未加载
评论 #26217437 未加载
评论 #26220901 未加载
评论 #26217606 未加载
评论 #26217663 未加载
评论 #26217255 未加载
LesZedCBover 4 years ago
personally i find a lot of arguments against AGI coming any time soon couched in a culture of human exceptionalism, even those who wouldn&#x27;t claim as much directly.<p>there is a DAMN surprising level of intelligence in significantly less complex life. we are just so attached to intelligence as defined by human culture to call it as it is.
评论 #26217774 未加载
评论 #26217707 未加载
评论 #26217919 未加载
评论 #26217704 未加载
评论 #26218050 未加载
评论 #26217975 未加载
评论 #26220150 未加载
评论 #26218312 未加载
评论 #26217395 未加载
dwohnitmokover 4 years ago
I think the reasoning presented in this article generalizes pretty well to a refutation of most arguments involving proving the impossibility of some complex, ill-defined, &quot;I&#x27;ll know it when I see it&quot; kind of phenomenon via a tidy, small logical proof.<p>There&#x27;s a lot of ways that those kinds of complex phenomenon can be functionally equivalent to human observers, but can have different underlying mechanisms. These tidy logical proofs only ever cut off one extremely specific incarnation of that complex phenomenon rather than the entire equivalence class.
评论 #26218573 未加载
Isinlorover 4 years ago
BTW - We empirically do not need AGI for computational&#x2F;intelligence explosion.<p>Single virus has no chance of out computing and evading an immune system, but billions of billions of viruses can out compute and evade even human civilization as a whole.
评论 #26221407 未加载
dan-robertsonover 4 years ago
I find the complexity arguments convincing despite the fact that they have various weaknesses (that while a solution to some NP-hard problem would suffice, it isn’t necessary and a merely good solution or a solution to an easier subset of problems would be ok.) There’s two things to talk about:<p>1. The article seems to switch a bit between superhuman intelligence or performance at something (I find this reasonably plausible) and a runaway singularity where intelligence&#x2F;performance grows <i>exponentially</i> and the computer then moves to destroy us all. A few reasons I find the latter proposition implausible are that only one problem of many needs to be too hard for complexity to be an issue; it’s not clear how good exponentially better performance is (for something like image recognition, how much does it matter to be able to recognise lots more things, or to have a 1% error rate that halves every year? I say not <i>that</i> much,) indeed it seems that evolution isn’t particularly optimising for intelligence in humans despite how much some of us may think it matters, although that could just be due to it not mattering so much until recently; and that it often relies on imagining an intelligence which is somehow clever in the way that humans and computers are clever (eg also fast at big mathematical calculations), but also stupid in the ways computers can be stupid, being precisely logical and optimising for a single goal (I find it unbelievable that the all-consuming paperclip-making intelligence would perfectly follow its paperclip-making instruction while destroying its masters and never thinking to disobey.)<p>2. I feel like most people who are serious believers in a singularity (eg Yudkowsky et al) are massive cranks who’ve reasoned themselves into a weird sci-fi corner. I struggle to believe that an argument from them is grounded in reality rather than begging their prophecy of a singularity future.
rpiguyshyover 4 years ago
each problem is different. computational chemistry has stagnated therefore AI isnt a concern? its nonsense. first of all, it may be that computational chemistry is much more tractable than we realize because we are too stupid to find the necessary footholds. but regardless, some tasks are actually mathematically intractable. there is no way to draw a connection between AI and any other problem, certainly not a connection definitive enough to write off the risk of AI...<p>that is the key. its all speculation. as long as there is some possibility of creating AI, we have to account for it in our collective decision-making. like many people before them, most people seem happy to write off the possibility of anything that hasnt happened already. fools.
SubiculumCodeover 4 years ago
I do wonder how far the human brain is from theoretical optimums. There is obviously something working really well, but there is also a lot of baggage that I do not doubt limits performance in some domains (cognitive) in order to preserve other basic functions (fight or flight or f*k). The biggest opponent to progress is ourselves. Even if you came up with an implant that would make humans smarter and more moral&#x2F;ethical, people won&#x27;t adopt it readily for fear of change. Unless AI becomes attached conservationist in nature, I&#x27;m not sure they&#x27;d retain the baggage not optimal for modern environments.
评论 #26220223 未加载
carapaceover 4 years ago
One problem with Singularity is that you either A) have to be first, or B) have to contend with other beings at least as intelligent as you are.<p>How can you be sure you&#x27;re first?
Inufuover 4 years ago
The author keeps referring to the &quot;PSPACE&quot; complexity of chess and Go in the context of AlphaGo; this is incorrect - these games are only PSPACE for arbitrary large board size N, at fixed board size as actually used by humans and current AI they are just constant O(1), complexity class is not relevant for this.<p>The article was also written before the best evidence we currently have for this was published: scaling laws for natural language understanding (<a href="https:&#x2F;&#x2F;arxiv.org&#x2F;abs&#x2F;2001.08361" rel="nofollow">https:&#x2F;&#x2F;arxiv.org&#x2F;abs&#x2F;2001.08361</a>), performance of RL algorithms with respect to data (eg AlphaGo Elo vs training time), image model accuracies, etc all show that exponentially increasing amounts of data&#x2F;computation are required for linear improvements in performance.<p>I posted some graphs with more details here: <a href="http:&#x2F;&#x2F;www.furidamu.org&#x2F;blog&#x2F;2020&#x2F;05&#x2F;03&#x2F;the-case-against-the-singularity&#x2F;" rel="nofollow">http:&#x2F;&#x2F;www.furidamu.org&#x2F;blog&#x2F;2020&#x2F;05&#x2F;03&#x2F;the-case-against-the...</a><p>tl;dr: current evidence suggests AI performance scales with log of data or computation
ytersover 4 years ago
There are no mathematical theories of runaway intelligence growth. On the other hand there are many theorems of fundamental limits to maechanical processes. E.g. NP completeness codiscoverer Leonid Levin also proved what he calls independence conservation that states no stochastic process is expected to increase net mutual information. Then there are the more well known theorems with similar implications: no free lunch theorems, halting problem, Kolmogorov complexity&#x27;s uncomputability, data processing inequality, and so on. There is absolutely nothing that looks like runaway intelligence explosion in theoretical computer science. The closest attempt I have seen in Kauffman&#x27;s analysis of NK problems, but there he finds similar limitations, except with low K terrains, but that analysis is a bit questionable in mind. To make arguments like gwern and Kurzweil they are essentially appealing to mysticism; assuming there is a yet to be discovered mathematical law utterly unlike anything we have ever discovered. They are engaging in promissory computer science, writing a whole bunch of theory checks they hope will be cashed in the future.
评论 #26217994 未加载
wwww4allover 4 years ago
Humans can already create intelligence. Called human babies.<p>Human babies are already nurtured, educated and developed into intelligent beings.<p>AI is like alchemy, trying to create something of value from nothing. People pontificating about AI is like medieval monks pontificating about how many angels can fit on pin head.<p>What is AI? What are boundary conditions of AI? Calling faster computers AI doesn’t make it sound more interesting.
fpgaminerover 4 years ago
We already have an existence proof for the singularity, so I don&#x27;t know why there&#x27;s any debate about _if_ the singularity will occur. I can see debate about what exactly the &quot;singularity&quot; entails, when, how, etc. But it&#x27;s inevitable.<p>The Cosmic Calendar (<a href="https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;Cosmic_Calendar" rel="nofollow">https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;Cosmic_Calendar</a>) makes it visually clear that progress is accelerating. Evolution always stands on the shoulders of giants, working not to improve things linearly, but exponentially.<p>When sexual reproduction emerged, it built on top of billions of years of asexual evolution. It took advantage of the fact that we had a set of robust genes. Now those genes could be quickly reshuffled to rapidly experiment and adapt on a time scale several orders of magnitude shorter than it would take asexual reproduction to perform the same adaptations.<p>Then neurons emerged; now adaptation was on the order of fractions of a life time rather than generations.<p>Then consciousness emerged. Now not only can humans adapt on the order of _days_, we can also augment our own intelligence. Modern day humans have access to the internet augmentation, giving us the collective knowledge of all humanity in _seconds_.<p>While we can augment our intelligence, the thing we can&#x27;t do is intelligently modify our own hardware. This is where AI comes in. With a sufficiently intelligent AI we could task it to do AI research for us. Etc, etc. =&gt; Singularity.<p>The vast majority of the steps towards Singularity have _already_ happened! Every step is an exponential leap in &quot;intelligence&quot;, and it causes adaptions to occur on exponentially decreasing time scales.<p>But I guess we&#x27;ll see for sure soon. GPT-human is a mere 20 years away (or less). I don&#x27;t personally think the AI revolution will be as dramatic as many envision it to be. It&#x27;s more likely to be like the emergence of cell phones. Cell phones undeniably changed and advanced the world, but it&#x27;s not like there was a single moment when they suddenly popped into existence and then from that point on everything was different. It&#x27;s hard to even point to exactly when cell phones changed the world. Was it when they were invented? Was it when they shrunk to the size of a handheld blender? When we had them in cars? The first flip phone? The first iPhone? The first Android?<p>The rise of AI won&#x27;t be a cataclysmic event where SkyNet just poofs into existence and wipes out humanity. It&#x27;ll be a slow, steady gradient of AI getting better and better, taking over more and more tasks. At the same time humanity will adapt and integrate with our new tool. When the AI gets smarter than us and hits the Singularity treadmill, we won&#x27;t just poof out of existence. More likely humanity, as a civilization, will just get absorbed and extended by our AI counterparts. They&#x27;ll carry the torch of humanity forward. They&#x27;ll _be_ humanity. Our fleshy counterparts won&#x27;t be wiped out; they&#x27;ll be an obsolete relic of humanity&#x27;s past.<p>More concretely, in 20 years we&#x27;ll have GPT-human, not as an independent, conscious, thinking machine. It&#x27;ll be a human level intelligence, but one bounded by the confines of the API calls we use to drive its process. That&#x27;s not something that&#x27;s going to &quot;wake up&quot; and wipe us out. It&#x27;s something we can unleash on our most demanding scientific tasks. Protein folding, gene editing, physics, the development of quantum computing. All being absolutely CRUSHED by an AI with the thinking power of Einstein, but no consciousness or the cruft of driving a biological body. It&#x27;s easy to see how that will change the world, but won&#x27;t immediately lead to humanity being replaced by free-willed AIs.
评论 #26218032 未加载
zxcvbn4038over 4 years ago
What if AI just wants to watch Star Trek reruns and browse Porn Hub? As is so often the case when humanity creates intelligences.
评论 #26216908 未加载
评论 #26216896 未加载
ProfHewittover 4 years ago
For state of the art on foundations of mathematics, see<p>&quot;Recrafting Foundations of Mathematics&quot;<p><a href="https:&#x2F;&#x2F;papers.ssrn.com&#x2F;abstract=3603021" rel="nofollow">https:&#x2F;&#x2F;papers.ssrn.com&#x2F;abstract=3603021</a>
评论 #26217808 未加载
评论 #26218022 未加载
coldteaover 4 years ago
The biggest pile of hand-waving I&#x27;ve seen...
评论 #26217003 未加载