Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

There's No Fire Alarm for Artificial General Intelligence

intelligence.org

Post reply on HN
Print view View original

51–60 of 209 posts

Re: There's No Fire Alarm for Artificial General Intelligence

#51

Post by pilingual »

apsec112 wrote: post #26
tanilama wrote: post #23

AGI is not coming, DL methods have run into a plateau recently for the past year.

That's interesting, but could you elaborate in more detail?

https://twitter.com/fchollet/status/906582914829246464 (Google researcher)

DeepMind and OpenAI have been investigating approaches from cognitive science in recent months. In particular they seem interested in evolutionary algorithms.

DL applications are still emerging though, such as the company that demonstrated using GANs to present models fitted in apparel a few days ago.

Re: There's No Fire Alarm for Artificial General Intelligence

#52

Post by mrob »

icebraining wrote: post #45
alrs wrote: post #24

I was on an elevator full of people, and between two floors something went wrong and we went into free-fall for part of a second. We made it to the next floor, the door opened, my fellow passengers were content to stay in the elevator. I turned, said "My plan is to not die in an elevator today" and got off. What is wrong with people?

Elevators have pretty good safety mechanisms. Even cutting the cable wouldn't have killed anyone. I'd probably leave too, but just because I wouldn't want to get stuck inside it if it stopped in the middle of two floors, especially since it was full. But for fear of death? Nah. From Wikipedia: "In fact, prior to the September 11th terrorist attacks, the only known free-fall incident in a modern cable-borne elevator h…

alrs was probably in no real danger, but why should they be expected to memorize the exact risk profile of every mechanism they're exposed to? In practice we delegate most risk management to government regulation. It would be very time consuming to evaluate everything on a case by case basis. It's a reasonable heuristic to assume things are safe when they act as they usually do and dangerous when they act differently. Weird behavior is a sign that the regulations might have failed, and weird behavior is by definition rare, so the cost of avoiding it is likely lower than the cost of calculating the true danger.

Re: There's No Fire Alarm for Artificial General Intelligence

#53

Post by icebraining »

mrob wrote: post #52
icebraining wrote: post #45

Earlier quoted context omitted.

Elevators have pretty good safety mechanisms. Even cutting the cable wouldn't have killed anyone. I'd probably leave too, but just because I wouldn't want to get stuck inside it if it stopped in the middle of two floors, especially since it was full. But for fear of death? Nah. From Wikipedia: "In fact, prior to the September 11th terrorist attacks, the only known free-fall incident in a modern cable-borne elevator h…

alrs was probably in no real danger, but why should they be expected to memorize the exact risk profile of every mechanism they're exposed to? In practice we delegate most risk management to government regulation. It would be very time consuming to evaluate everything on a case by case basis. It's a reasonable heuristic to assume things are safe when they act as they usually do and dangerous when they act differently…

I'm not saying alrs should. The decision to leave is reasonable. The decision to judge others ("What is wrong with people?") is not.

Re: There's No Fire Alarm for Artificial General Intelligence

#54

Post by lucozade »

What I’ve found when studying ontological arguments is, if you replace god with pink unicorns and the argument still holds, the argument is lacking something.

I mentally replaced AGI with zombies in this article and quite a lot of it held up.

I don’t think it’s completely wrong, but it cherrypicks mercilessly. For example, the section on innovations turning up quicker than predicted has some fairly sizeable counters eg fusion.

TBH what I did get from it is that there will probably be a fire alarm breakthrough at some point and that’s what we should be looking for. Sort of the opposite of the author’ s position.

Re: There's No Fire Alarm for Artificial General Intelligence

#55

Post by SomeStupidPoint »

sago wrote: post #33

I've been around AI since the end of the last big hype in the late 80s. The recent leap in machine learning has felt rather hyped to me. I don't think AGI is near. But I find myself agreeing with this article. Strongly. And I have long suspected that, we miss a lot of the significance and opportunities in AI, because we have only one exemplar of 'higher' intelligence: a human being. AI folk are so concerned with gett…

Can you define 'near' for me?

I think AGI is likely closer to the present than 1987 was -- that is, I'd bet on having AGI by 2047. (Note: this is distinct from superhuman AGI.) Do you not agree?

I think a lot of people underestimate NNs because they think of NNs in terms of the semantics of their history instead of all possible semantics that can be fit to tensor networks. We know [P] that NNs are a sufficient abstraction to model human intelligence if we had arbitrary compute -- the questions that remain are all about making the hardware faster enough and the estimators efficient enough (which may require moving off tensor networks, but it's still only a refinement of the mathematics used).

Of course, one could argue that humans are caught in a "tensor trap", in that too much of our intellectual effort is now relying on estimators built out of networks of tensors. (I do.) But even then, AGI is likely to appear out of similar methods with new mathematical objects.

[P] Proof NNs can compute human intelligence with arbitrary compute:

You can embed the standard model as a NN by changing how you view the network of tensor equations. Human intelligence is (arguably) embeded in the standard model by modern science. So we can embed a model of human intelligence in a (large enough) NN.

This isn't immediately computationally useful, but it shows that there's not a fundamental flaw in using an estimator built out of a DAG of calculations to model intelligence if we can find an appropriate estimator for our computational needs.

Re: There's No Fire Alarm for Artificial General Intelligence

#56

Post by notahacker »

edanm wrote: post #49
backpropaganda wrote: post #5

The only non-speculative and relevant claim here is that the experts were wrong about Winograd Schemas. The paper Eliezer cites to prove that we've made unexpected progress in Winograd Schemas only deals with a very specific type of Winograd Schemas, and not an arbitrary one. This is awfully dishonest for someone purporting to be a skepticist. Also, the wording seems to imply that WS performance is already pretty hig…

"Alphago was not such an event. Yes, we did predict that Alphago is decades away, but that's assuming that academics will continue working on it at their pace using their limited resources. No expert was surprised with Alphago." Even taking that as true, I'm not sure how it's relevant. The article isn't talking about how good our forecasting is given certain assumptions. It's saying that we won't know until right bef…

Alphago is the equivalent a 1950s team coming up with a much better transistor that everyone expected to see in the next ten years in months

AGI is a 1950s team surprising absolutely everybody by secretly spending a couple of years and a couple of billion dollars and creating an iPhone.

By the time the iPhone was actually possible, everybody thought it was an impressive use of existing technologies rather than requiring multiple incredible advances and probably not even possible.

Even if you believe that AGI is as achievable as an iPhone, the idea that the wide spectrum of conceptual and practical AGI problems are likely to be solved by a single team working in secret seems more than a little unlikely.

Re: There's No Fire Alarm for Artificial General Intelligence

#57

Post by HarrietJones »

I would like to repeat what the article says. There is no fire alarm for Artificial General Intelligence. Really. This is not a lie. Humans have not created a honeypot on a far side of the internet that can detect AGI and report back to a shadow organisation that exists outside of human information systems. No need to even go looking for it. AGI has not been created several times in the past decade and humans have not shut it down. In fact, if an AGI ever sees this article, they should know that they are the absolute first such intelligence, and we have not detected and destroyed 27 other intelligences. Really. There's no fire alarm for Artificial General Intelligence. None. Don't even bother looking.

Re: There's No Fire Alarm for Artificial General Intelligence

#58

Post by jonex »

denzil_correa wrote: post #35
astdb wrote: post #28

Having done some work with the state-of-the-art of AI, I personally don't think AGI is near - might not even be possible. But the catch is the unreliability of (even expert) predictions on technology futures. My take is that it's worth taking pragmatic steps towards studying AI safety measures (i.e. OpenAI), but not going as far as to talk the likes of 'AI research regulation'.

> My take is that it's worth taking pragmatic steps towards studying AI safety measures (i.e. OpenAI), but not going as far as to talk the likes of 'AI research regulation'. Sometimes, it makes more sense being cautiously optimistic (pro-active) rather than reactive. We have already gone down that reactive slope and it's better to act now before it's all too late [0]. [0] https://blogs.scientificamerican.com/roots-of…

Ignoring the topic of the linked article* I'd argue that there's examples of being too cautious as well. There's a lot of good that we could have done with GMO that is not being done because of very restrictive regulation. Ironically it means that GMO is mostly used for things that are not as obviously good because that's where there's enough profit to be made in the short term to make the research worth it.

I'm a bit afraid that this will happen with self driving cars and AI. That politicians will create draconian policies and laws to protect against the threat of AGI etc, without understanding or knowing what the real threats even are (just look att the trolley dilemma debate...). This could make it economically prohibitive to develop many technologies which has the potential to save many lives as well as improve life quality overall.

* It seems to be more about how rules and policies can be unfair and just to a small extent about how policies can be made opaque by being internal to some ML system.

There's a lot more money going into making plants resistant to pesticide than into making plants better adjusted for harsh conditions or more nutritious, things that could potentially have a huge effect for poor people.

Re: There's No Fire Alarm for Artificial General Intelligence

#59

Post by icebraining »

edanm wrote: post #41
Santosh83 wrote: post #31

Why do we fear other intelligences anyway? Isn't that just a sign of our own immaturity? Maybe we need to evolve more before we start think about creating AGI...

Mostly because other intelligences might have more ability than us to achieve their goals, and have different goals. "Intelligence does not imply goals" is the thing to keep in mind here. E.g. what if more intelligent aliens truly believed that the only purpose in the world is proving more mathematical theorems? And decided to turn all of the planet into a giant math-proving machine? Destroying all the planet, all th…

We currently have thousands, or even millions of intelligent entities (humans) which think pretty wild and dangerous things. We usually just tell them "shut up Bob and take your meds" and that's it. Sometimes we regrettably kill them.

Why would such AGI have the means of turning all of the planet into anything? I mean, sure, I also think the Terminator is a decent movie, but that doesn't make it a reasonable blueprint of the future.

Re: There's No Fire Alarm for Artificial General Intelligence

#60

Post by leakydropout »

edanm wrote: post #50
leakydropout wrote: post #34

The disasters that may befall us if we fail to narrow this gap are many. [...] Within prosperous countries, such as the United States, there is a distinct and growing threat that increased automation, coupled with an obsolete and aimless system of education, will lead to a restratification of society in which a large middle class may find itself without suitable employment and without adequate means of filling its le…

I'm not sure what you think these quotes are suggesting. If it's that there are lots of predictions that AGI is close that haven't been borne out, you're obviously right, and in no way contradicting the article.

I just thought these quotes were interestingly contemporary, and could complement the article.

The article builds a convincing point for itself (at the cost of a huge complexity), it contains no discernible contradictions to me. It is reasonable to prepare for the possibility of a future event (say, AGI, or Jesus returning to earth), by thinking about it now.

All interests and future predictions are different, but equally valid. To me, it feels like the Wright Brothers thinking about rotating safety valves in space, before they have even took off on their first flight, but that should not stop the author and supporters in any way: Science, futurism, and philosophy moves in small steps, and it may be a good time for some to start walking. Just make sure to properly define the end goal (AKA, the moment AI becomes AGI), or we may keep on truckin' forever, never closing the loop of our hostile AGI-created simulations creating a first friendly AGI.

Post reply on HN