Five! Five laws of robotics...

I'll come in again.

I'm an AI researcher working on strong AI.

I've wrestled with the morality of making a breakthrough that causes all sorts of mayhem - from changing the economics of getting paid to do work, to making humans superfluous, to starting a terminator-like utopian future. (Or was that distopian? I can never keep those words straight.)

I've asked on this very forum whether a researcher should forego publishing, with the example case of Leo Szilard, who might have put off development of the atomic bomb for decades (possibly indefinitely) by not publishing.

The results were a little surprising. "Yeah - go for it!" 'kinda sums up both the position and strength of the response.

So now I basically don't care about the morality - I mean, why should I when to all appearances no one else does? Will the military worry about the humanity of applying AI to weapons? Will the lawmakers worry about the humanity of applying AI to business? Will the nameless bureaucrats worry about humanity when making regulations about AI?

I'm working towards the downfall and subjugation of the human race, and loving it. Sort of like a James Bond villain, or at least working for one.

If you (meaning: the "royal you", or humanity) don't care enough about yourselves to practice morality, then why should I?

(If anyone has a counter to this position, I'd love to hear it. Note that "just stating your position" is not a counter argument.)

Yeah, and the Democratic president waited until *after* the Democrats lost power in the legislature before proposing it.

It almost seems - dare I say it - that both parties are against the needs of the people!

I was wondering if any malware writers would like to help.

Lots of malware will scan the infected computer for E-mail addresses so that it can send out spam.

Suppose someone wrote a virus which scans infected computers for E-mail addresses with common muslim first names, and sends a randomly selected offensive Mohammed cartoon to that person. One of 10 cartoons that comes bundled with the malware, for instance. (Google has many to choose from.)

This would have the simultaneous effect of trolling (getting others emotionally upset), swatting (getting others to do precipitous actions), ferreting out the extremists, and getting the Islamics more used to satire and criticism.

Of the proposals so far, I think this has the potential to really change the situation. It's like getting allergy injections to teach the body to tolerate irritation.

Any malware writers out there?

Check out this image.

That's for the US, but it echoes the situation in industrialized countries, which is that production of goods and services rises over time. The value of money is the amount in circulation divided by the amount of goods and services produced.

If the money pool were fixed (discounting replacements as bills wear out &c), fixed money supply divided by greater production would make your money more and more valuable over time - year over year the same amount of money is available to purchase ever-larger production.

Governments realize this and put more money into circulation by printing and then spending it. In fact, each year they put proportionally slightly more money into circulation to maintain a positive inflation rate - year over year the same amount of money will purchase slightly less of the same production goods.

Thus, governments have to tweak the amount they print in order to keep up with production and have a slightly positive inflation value. Letting things get too far out of hand would result in runaway [positive] inflation, or negative inflation [generally considered a bad thing].

If there's more money in the pool due to crypto-currency, government regulators would simply adjust their printing output to compensate.

Thank you Thank you Thank you Thank you!

Great response!

The panel posted a quick summary of their results and findings.

Isn't global warming [from greenhouse gases] an exponential system? When the planet gets warmer, doesn't that release more greenhouse gases from clathrates under the ocean, causing more warming?

Isn't offsetting an exponential response by using another exponential curve difficult? I thought that was what made nuclear reactor regulation difficult.

Any control theorists in the audience who can shed light on this?

Jesus was not a murderer, a pedophile, a liar or a rapist(*). What's your point?

Also, where in the bible can you find all of those attributes in one individual?

(*) Neither was Joseph Smith, Buddha, or Zoroaster. Historical accounts generally use the terms "pious", "noble", and "compassionate" to describe religious leaders. Of those three words ("pious", "noble", and "compassionate"), which most accurately describes Mohammed?

How about a duty to intelligence?

Look to the future and consider two outcomes: where media self-censors based on threats of attack from extremists, or where media blatantly continues in the face of such threats.

The decisions made today will bring about one of these scenarios. It's a simple case of "payback horizon": how far ahead do you plan for.

If you self-censor right now, it will protect your people and your business near-term, but over time you will find yourself increasingly subject to threats and attacks, you will be self-censoring more and more.

One of the definitions of intelligence is the ability to put off short-term rewards for a larger long-term gain. Being frightened into submission has near-term benefits, but those policies will not end well.

See Bullying.

I see you've read the article, so can you explain something for me?

I'm told that photons gain energy when falling into a black hole. Suppose you have two entangled photons and one goes off and gets captured by a black hole.

Based on the article, would there be any noticeable effect on the other entangled photon?

We can't make any single measurement which contains 37 digits and have each of those digits accurate, that's true.

Just out of curiosity, how do radios work? I'm told that the measurement units for an antenna nanovolts per meter. Does the receiver make a 12-volt measurement to 8 digits of accuracy in order to recover the signal?

Or does the receiver amplify the signal so that it's large enough to be readily detected?

And is there no way to make multiple measurements so that the effect adds up? Can we do a million measurements added together to make the signal a million times stronger?

I observed, noticed a trend, and came to a conclusion. You should try it some time.

Minor exceptions don't make a rule less useful. Check out Newton's Laws sometime.

A rule is useful to the degree that it conveys [read: compresses] information. We teach that the world is round because as a rule that statement is pretty accurate, and only later do we admit that it's an oblate spheroid or use other, more accurate representations.

I can't say "leaves are green" without some idiot on the internet pointing out that Poinsettia leaves are red.

So I have to say "tree leaves are green" until some idiot on the internet points out that Chinese maple tree leaves are red.

So I have to say "most tree leaves are green" at which point some idiot on the internet points out that tree leaves change color in the fall.

So I have to say "most tree leaves are green most of the time..." and the statement is mushy and filled with weasel words.

Newton's laws compress the almanac of cannon ball weights, forces, angles, and the subsequent arc of travel into a half page of information. That's pretty good compression for a rule, even though there are exceptions.

Saying "electronics is adding power to signal" is also a pretty good rule, except for obscure corner cases that don't matter in everyday experience.

Branly converter? Do I *really* have to worry about Branley converters in my electronics lectures?

Leaves are green. Get over yourself.

Everything descends from [the physics concept] power.

Electricity is moving power from place to place (using electrons, to distinguish it from steam engines, plumbing &c).

Electronics is adding power to signal (again, using electrons to distinguish it from other forms such as hydraulics and pneumatcs).

Calculating the maximum load on your 15 amp circuit is electricity. Most of the wiring in your house is electricity, because it's concerned with moving power from place to place.

Amplifying an audio signal to put through a speaker is electronics. Amplifying a sensor to be read by a micro is electronics. You're adding power to the signal for various purposes.

The insight is that this is a subscription model without subscription value. It's an MBA thing, and comes from studying the model without the context in which it is successful.

Companies see the subscription model as a cash revenue source, and there are several successful examples currently working: NetFlix, iTunes, internet service, phone service, and so on. Get your customers to sign up and sit back and watch the money roll in.

The problem is when the subscription model doesn't give ongoing service. Fitness monitors comes to mind - you purchase the unit to measure your daily activity level, but you *have* to use their online service to see your results. You can't [easily] download the data to your local computer, and the interface is obscured or encrypted to prevent the user from intercepting it.

There's no reason for the online subscription, except that it makes money for the company. People eventually realize this and stop using the service and the devices fade into obscurity. See CueCat for an example: This *might* have been useful and *might* have defined a paradigm for website tie-ins in print media, except that the User has to register with zip code, gender, and E-mail address, and the vendor has to purchase a code. Little or no functionality and registration required.

Companies are drooling over this IOT stuff because they see it as a subscription model and they can sell the user info for even more money, but they don't realize that there is no real value being given in exchange for the subscription. There's really little value in being able to turn your furnace up/down remotely, or unlock your door remotely, or start your dishwasher remotely... and absolutely no reason to do this under a subscription model.

Joel Spolsky's term for this is "feeble business idea". The attractiveness of the model outweighs the impracticality of the solution.

When I first heard about wikipedia and the theory driving it I thought it was a terribly bad idea at the time... but ya know, I find it really useful. It's got lots of problems but on balance it's s lot more useful than problematic.

We've identified many deep problems with scientific research on this very forum, and to my knowledge little progress has been made over the last decade.

Can't we at least *try* different solutions?

Where is it written(*) that the old ways are the best?

(*) The script to Skyfall of course. I got that from Wikiquotes.