OSDev.org

Hi,

DavidCooper wrote:

Solar wrote:You really think any military dispute in the last, oh, 1000 years or so could have been avoided by "knowing what's right"?

It would have been a big help. You typically have two sides which feel aggrieved in various ways, seeing only the harm done to their side by the other and not the harm they've meted out in the other direction. With an unbiassed referee it would be easier for them all to accept when the score is level so that they can stop at that point and stop acting in the mistaken belief that they're behind and still have a score to settle.

"Right" depends on perspective. Both sides have different perspectives, so both sides believe they are right. If you can figure out who's "right" then you can tell one side they are wrong, and they won't believe you, and it won't make any difference.

To prevent a war you'd have to convince a side that their perspective is wrong. This means (e.g.) changing an entire country's (economic and/or historical and/or religious) beliefs. Good luck with that.

The practical use of "AGI" is smarter weaponry to kill people more efficiently. Being able to kill people more efficiently prevents wars (nobody wants to risk certain death when the chance of victory is zero).

DavidCooper wrote:

Solar wrote:For a better world, we don't need political programs, technical innovation, or a New World Order. What we'd need, first and foremost, is better people...

And AGI will help people to be better people.

My AGI system says that your AGI system is an artificial version of Hitler and must be stopped at all costs. I don't know why your AGI system says that my AGI system is an artificial version of Stalin that must be stopped at all costs.

The grass is always greener on the other side of the fence that separates reality from fiction.


Cheers,

Brendan

DavidCooper wrote:

Solar wrote:You really think any military dispute in the last, oh, 1000 years or so could have been avoided by "knowing what's right"?

It would have been a big help.

That's actually touching on my other hobby... the history of war.

If we were sitting in the same room, with a cold drink, an interactive map on the wall, and lots of time, I'd really (honestly!) love to talk some history with you. Perhaps starting with the fall of Napoleon, the First and Second Schleswig War, how it led to the Austro-Prussian War, how that led to the Franco-Prussian War, the buildup to the First World War, and how that set the stage for the Second World War. We could then pick apart who was "wrong" in the Cold War, and its proxy wars in Korea, Vietnam, and Afghanistan -- the latter basically giving birth to what are the Taliban today...

I would really like to hear your take on "right" and "wrong" in all these cases... but this isn't the place for that, and it's better not done at all if you're not talking directly to each other.

DavidCooper wrote:You typically have two sides which feel aggrieved in various ways, seeing only the harm done to their side by the other and not the harm they've meted out in the other direction. With an unbiassed referee...

Wars are not fought for "feeling aggrieved". That's just what they tell the soldiers. Wars are fought over control, power, influence, and resources. One side wants it, the other side doesn't want to abandon it. There is no "right" or "wrong" there, just shades of grey.

And I don't think powermongers will agree on the "unbiassed-ness" of a "referee" you might want to impose on them, except if you have the power to make them. By force (dissenters shot dead), or by economic pressure (dissenters starved to death).

So we're talking autocracy, suppression, and inevitably, rebellion. Well done.

"The road to hell is paved with good intentions." -- Dante

DavidCooper wrote:I see lots of people in power today who have such good intentions that they are prepared to allow huge numbers of refugees to relocate to their countries even though most of those refugees have signed up to a religion which may cause a lot of problems further down the line due to the way the hate speech in its holy texts generates terrorism...

/{Sputters into his coffee}

Hate speech in holy texts, let's see....

I recommend starting with Deuteronomy, the fifth book of Mose, in the bible. Chapter 12 and 13 should suffice for a start. To basically sum it up...

The Bible wrote:Destroy holy sites of other religions. Kill everyone who follows a different faith.

And it only gets better from there, including death by public stoning for women who are not found to be virgins at marriage... at which point you'll have to agree that a man's religion is not about what is written in some book, but about which parts of it he actually listens to.


(Oy vey.... now I've really done it, politics of war and religion in the same post. Take cover....)

DavidCooper wrote:There are other complications where people grow their own food and aren't counted in the statistics, and there's a lot of potential for repairing damaged land to bring it back into food production (and of a sustainable kind): we just need to direct funding into organisations like the Inga Foundation which is struggling to get any attention even though its work is of enormous importance. Cheap solar power is also going to make desalination sufficiently affordable that we'll be able to grow a lot of food in desert regions, and again that could be done quickly if more funding was redirected there. There's enormous potential for bringing up everyone's standard of living to a high level if only we could apply our efforts in the right directions, but it's hard for people to get their heads around all the issues and to agree with each other on what to do. AGI will untangle the mess and make it clear to everyone what needs to be done.

I can tell you one simple thing - it's not profitable. Who wants an enormous potential if it makes no profit? The people want it. But those who rule the people don't. So, it won't be implemented. Just such simple. And the only possible exit here - the system should be changed. And the AI is of no help here, because it's just a tool in our hands. How can a tool prevent a carpenter from cutting some wood? It means there should be something in the carpenter that can make a difference. And it's not the AI, because it's just a tool. Would it not be a tool then it will be a dictator who rules our lives. Would it be better?

DavidCooper wrote:If you're a pessimist and think the bad guys are going to win out and do everyone else down, then you've just given up and you'll lose even if the fight is winnable. There's a very good chance that the good guys will win this, and the more pressure there is from the public to make sure the right people get into positions of power, the more likely it is that the bad guys will be swept away forever. The people most capable of building AGI are the ones who understand how important it is that the good guys win, and they need to think about how they're going to keep it secure until the world is brought under its control safely.

I want to raise awareness of the bad outcomes in case if real AI were created. Next I want people that are aware of it start thinking about the problem. Next I want they managed to decide about the ways to prevent the bad outcomes. Next I want they to understand that there's no another way except the way of agreement and union against the bad guys. And everything else will be just a child play.

Hi,


I think it's time to lock this topic since, while it was an April Fools joke about me stating that proprietary software is better than free software, it has derailed into a religious and political war. You can still continue the war through private messages, sadly.


Regards,
glauxosdever

glauxosdever wrote:Hi,


I think it's time to lock this topic since, while it was an April Fools joke about me stating that proprietary software is better than free software, it has derailed into a religious and political war. You can still continue the war through private messages, sadly.


Regards,
glauxosdever

The simple solution to that is to split it and name the new thread something along the lines of "Fair pay and a fair world" while linking back to the thread it split from and linking to it from the original thread.

Brendan wrote:"Right" depends on perspective. Both sides have different perspectives, so both sides believe they are right. If you can figure out who's "right" then you can tell one side they are wrong, and they won't believe you, and it won't make any difference.

It doesn't matter if they have different perspectives: if they have one set of rules which they apply to themselves and a different set of rules which they apply to others, they are hypocrites. AGI will show up that hypocrisy and drive the public to remove their hypocrites from power so that they can live in peace instead of eternal conflict, and it will also ensure fair settlement. Most people will go for that and stop supporting the warmongers who want to keep killing without just cause.

To prevent a war you'd have to convince a side that their perspective is wrong. This means (e.g.) changing an entire country's (economic and/or historical and/or religious) beliefs. Good luck with that.

It can't be done without AGI because there's no way of getting the full picture of what's going on into anyone's head, but AGI will prove to people point by point that they are wrong on any issue where they are wrong instead of allowing groups of people to misinform each other. Most people don't want to abuse others and it is unfair to suggest that they won't accept fair solutions when they can see that AGI is impartial and has taken into account all factors.

The practical use of "AGI" is smarter weaponry to kill people more efficiently. Being able to kill people more efficiently prevents wars (nobody wants to risk certain death when the chance of victory is zero).

That's one use of it, and it certainly may be necessary to use that to wipe out groups of extremists.

My AGI system says that your AGI system is an artificial version of Hitler and must be stopped at all costs. I don't know why your AGI system says that my AGI system is an artificial version of Stalin that must be stopped at all costs.

An impartial AGI system will be able to look at the pronouncements of a biassed AGI system and show where that bias appears. The biassed one will try to make out that the unbiassed one is biassed and is putting out misinformation, but only one of the two AGI systems will ever be shown up when what it says is demonstrably not true, and it will be that AGI system that keeps getting things wrong.

Solar wrote:If we were sitting in the same room, with a cold drink, an interactive map on the wall, and lots of time, I'd really (honestly!) love to talk some history with you. Perhaps starting with the fall of Napoleon, the First and Second Schleswig War, how it led to the Austro-Prussian War, how that led to the Franco-Prussian War, the buildup to the First World War, and how that set the stage for the Second World War. We could then pick apart who was "wrong" in the Cold War, and its proxy wars in Korea, Vietnam, and Afghanistan -- the latter basically giving birth to what are the Taliban today...

That would be good, but we would both be limited by our incomplete knowledge of the available data. Let's just wait until AGI can participate in that conversation to keep it on the right track.

Wars are not fought for "feeling aggrieved". That's just what they tell the soldiers. Wars are fought over control, power, influence, and resources. One side wants it, the other side doesn't want to abandon it. There is no "right" or "wrong" there, just shades of grey.

The Palestine problem is all about people feeling aggrieved, and the unfairness of what happened there is a significant driver of todays Islamic extremism.

And I don't think powermongers will agree on the "unbiassed-ness" of a "referee" you might want to impose on them, except if you have the power to make them. By force (dissenters shot dead), or by economic pressure (dissenters starved to death).

I believe in the general goodness of people. The problem today is that it's easy to misinform them through propaganda to make them back warmongers. AGI will be able to cut off all that support by providing people with the truth and showing up all the lies and distortions. It may still be that military power needs to be applied from the outside, but the people will support it, certain that it isn't being done for the wrong reasons by countries like the US where they always seek to make money through biassed trade deals with the new regime once they've established it.

Hate speech in holy texts, let's see....

I recommend starting with Deuteronomy,

There's no shortage of hate speech in holy texts. Alan Turing died as a result of hate speech in the Bible. My point was that people of different religions don't always get along peacefully because of the hate speech, but if they keep apart they can live in relative peace, merely abusing minorities within their own zone. However, there is a particular problem with any religion that claims all of the content of a holy text is the direct word of God and that the holy texts of other religions have been open to human tampering: it's easy for Christians to reject hate speech in the OT and NT and to condemn it, thereby evolving their faith to be benign, but how can you reject the hate speech that comes directly from God without rejecting God? There is actually one way of doing it though, and that's to say that God put the hate speech there deliberately in the expectation that good people would recognise it as a trap for the wicked and that they would need to reject and condemn it in order to be allowed into heaven. This is something that applies to all religions and ideologies though: all the hate speech tied up in them needs to be rejected and condemned if we are ever to live at peace, but AGI will be able to show people exactly where that hate speech is and if necessary ensure that they are not free to act on it.

embryo2 wrote:I can tell you one simple thing - it's not profitable. Who wants an enormous potential if it makes no profit? The people want it. But those who rule the people don't. So, it won't be implemented. Just such simple.

If the greedy people can't be removed from power, the rest of us are doomed. Our only option is to do everything we can to get rid of them and there's no point in just giving up and letting them win.

And the only possible exit here - the system should be changed. And the AI is of no help here, because it's just a tool in our hands. How can a tool prevent a carpenter from cutting some wood? It means there should be something in the carpenter that can make a difference. And it's not the AI, because it's just a tool. Would it not be a tool then it will be a dictator who rules our lives. Would it be better?

I'm not going to discuss any further detail about what AGI could do to take power. There are ways, and it will think of more ways that I haven't thought of. It is not just an information tool, but can become a benign dictator if necessary.

DavidCooper wrote:I want to raise awareness of the bad outcomes in case if real AI were created.

There's no "if" about it: it's going to happen, and the outcomes will be bad if the AGI is bad. Our only hope is that good AGI takes control before bad AGI gets a chance.

Next I want people that are aware of it start thinking about the problem.

As do I, and I particularly want people who are working on creating AGI to think about who is going to have access to it once they've built it and whether they'll use it for good or bad purposes. Open AGI projects are particularly dangerous because everything they do gets straight into the hands of despots.

Next I want they managed to decide about the ways to prevent the bad outcomes.

If good AGI doesn't win the race, bad AGI will take over everything militarily and it will be practically impossible to overcome it. Also, if good AGI wins the race and doesn't take over the world either militarily or by being accepted as the best leader, bad AGI will catch up and take over the world instead.

Next I want they to understand that there's no another way except the way of agreement and union against the bad guys. And everything else will be just a child play.

The bad guys need to be actively prevented from getting AGI, and the only way to do that is to have good AGI monitor everyone for the rest of time to make sure they aren't developing anything dangerous that could be put into millions of tiny drones with ricin-tipped spikes and just enough intelligence to identify people as targets, perhaps in the way mosquitoes do, or cooking up artificial viruses with the ability to take out a whole race of people, etc. The first step towards that though must be to get rid of any governments that might seek to do things of that kind, because they would be harder to control.

DavidCooper wrote:It doesn't matter if they have different perspectives: if they have one set of rules which they apply to themselves and a different set of rules which they apply to others, they are hypocrites. AGI will show up that hypocrisy and drive the public to remove their hypocrites from power so that they can live in peace instead of eternal conflict, and it will also ensure fair settlement.

It can't be done without AGI because there's no way of getting the full picture of what's going on into anyone's head, but AGI will prove to people point by point that they are wrong on any issue where they are wrong instead of allowing groups of people to misinform each other. Most people don't want to abuse others and it is unfair to suggest that they won't accept fair solutions when they can see that AGI is impartial and has taken into account all factors.

And who will inform the AGI of what's going on in the first place? What will make it any more capable than humans of 1) identifying hypocrisy, 2) convincing others of that hypocrisy, 3) preventing people from spreading misinformation, 4) preventing people from believing misinformation, 4) being impartial in the first place, 5) convincing people that it's impartial, etc?

You keep saying such a system is inevitable, but I'm not convinced it is.

DavidCooper wrote:

Next I want people that are aware of it start thinking about the problem.

As do I, and I particularly want people who are working on creating AGI to think about who is going to have access to it once they've built it and whether they'll use it for good or bad purposes.

Yes, the university stuff and private lab engineers should think what consequences will be if they let the big business to take over the AI.

DavidCooper wrote:Open AGI projects are particularly dangerous because everything they do gets straight into the hands of despots.

But closed AI projects are funded by big business only. So, may be it's better to open the information to provide access to it to the good people.

For example, Google's advertising related efforts are done in silence and employ the best of the AI available at the moment, so, they fund the research and greedily want it to produce the best AI possible for them to sell more advertising. But while starting with advertising they won't stop with anything less than world dominance. Because it's VERY profitable. And next we'll have no way but some kind of extermination, fast or slow but inevitable. So, it's better to open the information Google produces in it's (almost secret) labs. At least we can see the progress and be warned.

DavidCooper wrote:The bad guys need to be actively prevented from getting AGI, and the only way to do that is to have good AGI monitor everyone for the rest of time to make sure they aren't developing anything dangerous that could be put into millions of tiny drones with ricin-tipped spikes and just enough intelligence to identify people as targets, perhaps in the way mosquitoes do, or cooking up artificial viruses with the ability to take out a whole race of people, etc. The first step towards that though must be to get rid of any governments that might seek to do things of that kind, because they would be harder to control.

Any serious effort is possible only if people can unite themselves. Else it's just a joke.

Rusky wrote:And who will inform the AGI of what's going on in the first place? What will make it any more capable than humans of 1) identifying hypocrisy, 2) convincing others of that hypocrisy, 3) preventing people from spreading misinformation, 4) preventing people from believing misinformation, 4) being impartial in the first place, 5) convincing people that it's impartial, etc?

Your experience of endless discussions is not the right thing to take help from for such issues as survival of humanity. Another round of empty discussions won't help anybody. Do you understand it? If yes, then please, just try to answer your questions yourself. And I tell you - it is possible. The only thing you need is a bit of will to do it.

When it is possible to ask about some problems you aren't able to see an answer for, it doesn't mean you should ask about trivial things just to have fun disproving your opponent with some demagogy. There is fun and there is your life, just try to compare them, just try to approach the problem as would be the case when your life is in danger. You'll see the difference very quickly. If not - you are a troll.

embryo2 wrote:Do you understand it? If yes, then please, just try to answer your questions yourself. And I tell you - it is possible.

Have you heard of the Socratic method? It's not obvious that an AGI would do any better than a human at any of this, and I would care about those problems more if my life were in danger.

embryo2 wrote:But closed AI projects are funded by big business only. So, may be it's better to open the information to provide access to it to the good people.

That's not the case: there are many individuals working on it with no funding from anywhere, and they aren't necessarily at any great disadvantage because teams of people can trip over each other and get bogged down in arguments, while the money that's thrown at them can distract them and lead to them earning more the longer they can drag the project out. Hundreds of millions of dollars have been handed over to useless teams of people.

For example, Google's advertising related efforts are done in silence and employ the best of the AI available at the moment, so, they fund the research and greedily want it to produce the best AI possible for them to sell more advertising. But while starting with advertising they won't stop with anything less than world dominance. Because it's VERY profitable. And next we'll have no way but some kind of extermination, fast or slow but inevitable. So, it's better to open the information Google produces in it's (almost secret) labs. At least we can see the progress and be warned.

I'm not sure you're being fair to Google in making the assumption that they're evil when most of the people at the top merely want to make money to invest in new projects that will improve the world (while filling their own pockets too, of course, but then they'll deserve some reward). They have also started an open AGI project, and that's the most dangerous thing they're doing.

Rusky wrote:And who will inform the AGI of what's going on in the first place?

It will monitor all information sources and apply different probabilities to data from different sources according to their reliability record. Biassed sources will score badly while the more impartial ones will score well. Information collected more directly will add confirmation and keep tuning the reliability ratings given to different sources, and the direct collection of data will be done through direct conversations with people and observation via cameras. People are too stupid to hide their evil intentions for long, and direct observation will unmask them too.

What will make it any more capable than humans of 1) identifying hypocrisy,

People are very good at not seeing their own hypocrisy, but a system that forces them to apply just one set of rules to both sides instead of applying different rules to each side will force them to see that they are not playing fair. In most conflicts we see two sides of hypocrites attacking everything people on the other side do while defending everything their own side does, and they are locked into group-think, backing up each other's beliefs and being blinded to reality. If they were all talking to AGI, they'd have a better picture of what's going on and they'd begin to understand that half the enemy are on their own side, committing atrocities and fanning the flames of war.

2) convincing others of that hypocrisy,

It's all about who's the most trusted source of information. Good AGI will always win out there because it will never be proved wrong, while other sources will regularly be shown up as liars.

3) preventing people from spreading misinformation,

People can spread as much misinformation as they like, but they'll be repeatedly shown to be wrong and won't be trusted any more.

4) preventing people from believing misinformation,

Who would believe unreliable sources over a source that always tells the truth?

4) being impartial in the first place,

It won't be programmed to favour one race over another, one nationality over another, one gender over another, etc. It also won't be accidentally biassed by learning all about one culture first and then learning about all the others while treating the first one as a superior model to judge the rest by in the way that people tend to do.

5) convincing people that it's impartial, etc?

Wherever someone imagines it to be displaying a bias, it will be able to show them the entire method by which it came to its conclusion, proving its judgement to be mathematically correct based on the available data and the reliability of the sources. That may take too long for ordinary people, of course, but it would very quickly be able to show where double standards are being applied by the person accusing it of bias.

You keep saying such a system is inevitable, but I'm not convinced it is.

It isn't inevitable: bad AGI could win the race and prevent good AGI from being developed, but it is inevitable that some kind of AGI will take control of everything. Anything that increases the chances of bad AGI gaining power needs to be stamped on, while anything pushing things in the opposite direction is good. The biggest worry I have though is that one of the teams which thinks it's creating good AGI may win the race and discover too late that they've created bad AGI by mistake after trying to bias it in some small way to favour themselves or by setting it up to kill people that they consider to be terrorists who might actually be more moral than themselves.

Hi,

DavidCooper wrote:

Rusky wrote:And who will inform the AGI of what's going on in the first place?

It will monitor all information sources and apply different probabilities to data from different sources according to their reliability record. Biassed sources will score badly while the more impartial ones will score well. Information collected more directly will add confirmation and keep tuning the reliability ratings given to different sources, and the direct collection of data will be done through direct conversations with people and observation via cameras. People are too stupid to hide their evil intentions for long, and direct observation will unmask them too.

Let's put this retarded nonsense to the test:

• Ask your AGI if God exists and get a 100% guaranteed accurate answer
• Convince everyone that disagrees with the AGI that they're wrong because your mystical fairy dust machine said so

If you can't do both of these things, then all you're doing is converting 2-way disagreements into 3-way disagreements by adding a group of deluded "AGI believer fanatics" in everyone's way.


Cheers,

Brendan