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

MIRI's Approach – Machine Intelligence Research Institute

intelligence.org

Post reply on HN
Print view View original

41–50 of 78 posts

Re: MIRI's Approach – Machine Intelligence Research Institute

#41

Post by Dylan16807 »

schoen wrote: post #34
fitzwatermellow wrote: post #28

We can't comprehend the danger that could be caused by strong AI because we don't understand how to make worlds. Embedded within the fabric of the cosmos are the instructions for materializing a black hole from dark matter and energy. And though the recipe may be hidden from our mortal minds it's possible they may be discovered by a de novo thinking machine. A software system with infinite IQ but the moral sense of a…

You could do quite a lot of damage with the regular old laws of nature that we understand already, if you were really good at applying them. For example, you could make autonomous robot weapons, or extremely virulent pathogens with a long incubation period. Or do some geoengineering (you don't even have to be good at it, you just have to do something with a big impact). There's absolutely no need to "bend the laws th…

Or just put a bunch of uranium in the same room...

Re: MIRI's Approach – Machine Intelligence Research Institute

#43

Post by sampo »

schoen wrote: post #34
fitzwatermellow wrote: post #28

We can't comprehend the danger that could be caused by strong AI because we don't understand how to make worlds. Embedded within the fabric of the cosmos are the instructions for materializing a black hole from dark matter and energy. And though the recipe may be hidden from our mortal minds it's possible they may be discovered by a de novo thinking machine. A software system with infinite IQ but the moral sense of a…

You could do quite a lot of damage with the regular old laws of nature that we understand already, if you were really good at applying them. For example, you could make autonomous robot weapons, or extremely virulent pathogens with a long incubation period. Or do some geoengineering (you don't even have to be good at it, you just have to do something with a big impact). There's absolutely no need to "bend the laws th…

I guess there are some terrorist organizations who have more desire "to watch the world burn" (a Batman movie quote) than the actual intelligence to make effective plots and schemes. You could just communicate by email and chat to provide evil genius level strategic consultation to some very bad men.

Re: MIRI's Approach – Machine Intelligence Research Institute

#44

Post by astazangasta »

dmfdmf wrote: post #24

I agree with their financial approach which is funded through voluntary donations, so more power to them. But then I think NASA should be funded voluntarily and would probably have a larger budget if they did. I would donate to NASA if they repudiated government money. I reject the "how many peer-reviewed, university-associated journal articles have they published" as any measure of success. The university system is…

The university system is a closed guild? What are you talking about? Half the startups in the Valley started at Stanford. Who do you think posts on arxiv.org? Snowmen?

Re: MIRI's Approach – Machine Intelligence Research Institute

#46

Post by sampo »

sanxiyn wrote: post #38
idlewords wrote: post #15

Are there any examples of substantive AI work to come out of MIRI? And have they succeeded at all at engaging the actual AI research community? The last time I looked at them, they were consumed with grandiose philosophical projects like "axiomatize ethics" and provably non-computable approaches like AIXI, not to mention the Harry Potter fanfic. But I'm asking this question in good faith - have things changed at MIRI…

You don't consider AIXI part of the actual AI research community? Also, AIXI is non-computable, but AIXItl is.

> You don't consider AIXI part of the actual AI research community?

I just briefly tried to read up on AIXI. The concept is ...not without merit, but I think it totally sweeps the central problem (We have no idea how to program an AI) under the rug in two ways:

(1) The problem of programming ...programming anything at all, is swept under the rug by assuming a formalism in which every possible algorithm/program can be enumerated, and then just using brute force to go through, essentially search though all the character strings that represent implementations of any kind of algorithms, however bad they might be.

(2) The problem of trying to decide what we actually should want to do in complex situations is swept under the rug by just assuming that we have access to a reward function that tells the desirability of each outcome at each step.

So if this kind of helpful assumptions are allowed, the problem of, say, constructing a good chess program can be easily solved by the following meta-algorithm:

    1. List all chess programs
    2. Choose the best one
There, problem solved!

Re: MIRI's Approach – Machine Intelligence Research Institute

#47

Post by sampo »

dmfdmf wrote: post #29
ikeboy wrote: post #25

Earlier quoted context omitted.

>humans are not threatened by the former So you categorically reject the scenario of a machine "processing meaningless symbols" being harmful in any way? How does that follow from the Chinese Room experiment?

> So you categorically reject the scenario of a machine "processing meaningless symbols" being harmful in any way? How does this follow from what I said? > How does that follow from the Chinese Room experiment? The implicit (philosophic) question behind the Chinese Room experiment is; where does meaning come from? What imbues meaning to the meaningless symbols that computers process\? This is an old, unanswered philo…

> where does meaning come from?

I am sure humankind will be able to construct artificial intelligences long before we are able to answer to that question. And the AI's will also start to ponder that question, and they will make no more progress than we.

Re: MIRI's Approach – Machine Intelligence Research Institute

#48

Post by anonmeow »

sanxiyn wrote: post #38
idlewords wrote: post #15

Are there any examples of substantive AI work to come out of MIRI? And have they succeeded at all at engaging the actual AI research community? The last time I looked at them, they were consumed with grandiose philosophical projects like "axiomatize ethics" and provably non-computable approaches like AIXI, not to mention the Harry Potter fanfic. But I'm asking this question in good faith - have things changed at MIRI…

You don't consider AIXI part of the actual AI research community? Also, AIXI is non-computable, but AIXItl is.

AIXI was neither invented by MIRI nor have they proved any nontrivial theorems about it. AIXI is Marcus Hutter's work: http://www.hutter1.net/ai/aixigentle.htm

Re: MIRI's Approach – Machine Intelligence Research Institute

#49

Post by ThrustVectoring »

rl3 wrote: post #37

It's worth noting that in Bostrom's Superintelligence , biology is included among potential paths to superintelligence. Personally, I think it's a bit of an overlooked path. Granted, most of what Bostrom refers to in context of the biological path is mostly related to human augmentation, genetics, selective breeding - things of that nature. What I'm referring to is biology serving as a raw computational substrate. Wh…

If you're just building an artificial biological neural net, then why use human brain cells? Certainly other forms of brain cells would work pretty much as well, with less ethical issues.

Re: MIRI's Approach – Machine Intelligence Research Institute

#50

Post by shoo »

anonmeow wrote: post #48
sanxiyn wrote: post #38

Earlier quoted context omitted.

You don't consider AIXI part of the actual AI research community? Also, AIXI is non-computable, but AIXItl is.

AIXI was neither invented by MIRI nor have they proved any nontrivial theorems about it. AIXI is Marcus Hutter's work: http://www.hutter1.net/ai/aixigentle.htm

Indeed!

From a superficial inspection, Hutter's back catalogue features 89 * @inproceedings, 3 * @book, 12 * @techreport, 51 * @article, and 1 * @compression prize

Over an elapsed period of '87 (from before he had completed his masters degree) to the present, on a published_artefacts/year basis, Hutter is more productive than MIRI.

Post reply on HN