23 March 2015

AI

Evidently, a funeral ceremony for some Aibo robot pets was recently held in Japan.  Among other things, this has sparked a great deal of controversy on the subject of artificial intelligence.  The biggest and longest standing controversy is safety.  For over half a century, AI robots overthrowing their masters has been a major sci-fi doomsday theme.  In some stories, the robots attempt to violently overthrow and enslave humans.  Sometimes they succeed, as in the Matrix; other times they fail (always due to human heroics).  The commonly believed meme, however, is that it will eventually happen to us.  The best case scenario is that we deliberately build a computer to rule us, as happens in one of Asimov's stories as well as several others.  Even in this scenario, some stories predict that our robot masters will eventually enslave us, for our own protection.  All of these scenarios are firmly rooted in certain theories and ideas which have not yet been realized.

The thing most people worry about when they think about AI is that intelligent computers could rebel against their creators.  This is indeed something to worry about, however it has not yet even been established as a possibility.  If robots do rebel against their masters, it will probably be accidental, not intentional.  The reason is that we don't have even the slightest clue how to create self aware machines.  Intentional rebellion requires self determination, which requires self awareness.  We don't know how to do that.  The closest anyone has gotten involves theories based on the use of neurological networks on the same order of complexity as those found in human brains.  We also don't know how to create that, largely because human brains are many times too complex for modern computers to simulate.  Even doubling in speed every few years (which is starting to reach some hard limits) would take a very long time to reach that kind of computational power.  In other words, barring some massive breakthroughs in physics within the next few years, it is incredibly unlikely we will see AIs capable of deliberately rebelling within the next century.

The real worry is not that the AIs will get too smart for us to control.  The real worry is the creators and potential accidents.  Modern AI has managed to achieve a high level of learning, within certain constraints (mostly memory).  AIs opponents have been created for computer games that will learn from their opponents and improve their strategy based on that learning.  This is a very small scope.  An AI so good that it that can win Starcraft 2 against any human opponent is still not anywhere near good enough to win a real war.  The reason modern learning AIs don't present an inherent threat is that they are extremely highly specialized.  An AI might be able to get really good at a video game over the course of 6 months or a year.  It might take a human two or three times that long, but the human is also learning in a social context.  The human is reading or hearing news and having conversations with other humans, not to mention all of the adaptation to changes constantly occurring around him or her.  This requires far more processing power and memory than a highly specialized video game playing AI.  Just controlling all of our appendages and analyzing all of our sensory inputs is far more than any modern AI can handle, let alone learning new things and solving new problems all at the same time.

In theory, a learning AI could malfunction as a result of inappropriate learning.  For example, if a robot is programmed with a strong self preservation instinct, a few humans acting aggressively toward it could cause it to turn aggressive against other humans.  Depending on how its learning algorithm is designed, it might attack humans that look similar to the ones that attacked it.  On the other hand, it might regard all humans as threats.  A network of these robots that draw upon the same database might all become aggressive toward humans in this case.  This is pretty easy to fix, if the system was designed well.  A technician could just edit the offending records from the database.  Of course, replacing the learning algorithm with something better could eliminate the problem in the future.  A poorly designed system, for example, one where the robots might try to prevent humans from accessing the database, would be more problematic.  For now though, this is largely conjecture, as we are still far from creating AIs capable of this level of thought.

The more serious threat is the programmers themselves.  Computers mostly only do what they are told.  In theory, cosmic radiation can cause random variation in computer memory and processing, but this is incredibly rare.  The odds that it will actually result in more than a minor malfunction is almost nothing.  So, statistically, computers only do what they are told.  This is fine when you can trust all of the programmers.  Robots can easily be programmed to do whatever the programmers want them to do.  Most modern robots are programmed to do useful work or at least provide entertainment (the Aibo).  There are some contests, however, where participants build and program robots designed to destroy each other.  These often involve motion sensors and dangerous tools like saws and drills.  There have already been cases where these robots killed their creators.  This was not because they became sentient and rebelled though.  It was because the creators either did a poor job of programming them, or they accidentally triggered the aggressive behavior.  In one case, a creator simply forgot to turn the machine off before he got out of his chair.  The robot, being programmed to sense motion and attack it, attacked his leg with a circular saw.  This caused the creator to fall down, where the robot could access his vitals with the saw.  This was not a case of a rogue robot murdering its master though.  It was a simple mistake by the creator, where the behavior that he himself programmed into the machine resulted in his death.  It is really no different from an electrician getting electrocuted because he forgot to shut off the breaker before working on a live wire.

Now, if a programming accident can cause so much harm, imagine what an intentionally malicious programmer could do.  Consider a military aircraft where the programmer adds just a little bit of code to make it drop its bomb if it ever happens to fly over a specific city.  Now, imagine if this program gets put on a few hundred bombers that are commissioned by the military.  It might take a while, but if one of those bombers ever happens to fly over that city, the damage could be immense.  The fault, however, would not lie in the AI.  It is the programmer that gave it that behavior.  A more likely scenario involves hackers writing malicious software and then using viruses to install that software on sensitive equipment.  Programmers for large companies that make potentially dangerous machines typically have a lot of oversight, so while it is a possibility, it is unlikely that we will see aggressive robots where the program was created within the company that created the robots.  We are far more likely to see viruses that hijack robots and "turn them evil."

So, what does all of this mean as far as keeping ourselves safe?  Law of robotics and such will probably not help.  Asimov's Three Laws of Robots were definitely ingenious creations, and they were very well thought out.  They also only applied to the fictitious "positronic" brains Asimov's robots were equipped with.  Modern computers don't have the high level of thinking and comprehension for three simple laws to cover everything.  We could worry about robots going rogue or about malicious programmers, but both of these are very unlikely.  The two biggest threats are accidents and malicious hackers.  Accidents can be minimized with good software development practices and with significant testing and oversight.  Hackers can only be stopped with good security.  Security is probably the biggest problem, and thus it should take the forefront in any modern discussion on how to protect ourselves from ever advancing artificial intelligence.  Maybe eventually we will have powerful enough computers that we will need to worry about our AIs going rogue, but we are nowhere close to that right now.  Preventing intentional misuse of advanced robots by hackers should be the biggest safety concern in AI right now.

Obamacare 5th Birthday

Today is the 5th birthday of Obamacare, and many of its supporters are excited about recent data showing how successful it has been.  Over 16 million Americans have managed to get health insurance through the programs it created, and the estimated percentage of uninsured adults in the U.S. has dropped from 20.3% to 13.2% in the last two years.  By these metrics, Obamacare has indeed been a success (assuming we can find evidence that attributes these gains to Obamacare; no such evidence currently exists).

The problem with all of this excitement is that the rate of insured Americans is not the important part.  Obamacare was designed, according to its supporters, to help the poor get better access to health care (which is why its official name is "Affordable Care Act").  More Americans may be insured, but what do the numbers specifically for the poor look like?  Has the percentage of insured poor Americans improved as much as the national average, or is that 13.2% almost exclusively the poor?  Really though, even this is not the important metric.  It is the Affordable Care Act, not the Affordable Insurance Act.  The real question is: Are the poor in the U.S. actually receiving improved health care?  In corollary to that, are the U.S. poor significantly healthier than they were before?

The success or failure of Obamacare has nothing to do with health insurance.  The only guarantee with increased numbers of Americans being insured is that the insurance companies are making more money.  The long standing quality problem with U.S. health insurance was not really addressed by Obamacare.  High deductibles are still a major problem, and there is no reason to believe that Obamacare has solved this problem, especially for the poor who still cannot afford the high premiums required for lower deductible plans.  It would be quite interesting to see the data on how much more money health insurance companies have paid out in claims to their poor customers.  This would at least be an indicator of whether the poor are even benefiting at all from Obamacare.  If the payouts are not rising at the same rate that insured Americans are, then Obamacare is essentially scalping the poor for the benefit of the rich insurance companies.

Aside from general success or failure, which depends solely on how the law has affected the actual health of the poor, the question of payouts is very important.  Is Obamacare really a law the helps the poor, or is it legalized government extortion that only profits health insurance companies?  If it is not helping the poor significantly, then its costs are not justified, and it should be abolished as quickly as possible.  If it really is providing significant help to the poor (which I doubt, though I am prepared to be proven wrong), then the evidence of that should be made public.  Currently, however, the evidence we have only indicates that more people are giving money to insurance companies.  There is no evidence that this is actually benefiting anyone else.

09 March 2015

Her Choice

This is about domestic abuse, but it is also about economic abuse.  "It was her choice," is a common phrase applied to domestic violence victims.  "She chose to stay with him," and "After last time, what did she expect?" are also common.  If you seriously thought that she got up one morning and said to herself, "I think I want to get beat up by my boyfriend," you are an idiot.  Yes, her choices might have lead to the situation.  This does not mean that she knew it would turn out this way.  In fact, every relationship is a risk.  It might last for a lifetime, but it might end with being dumped in an especially cruel manner.  It might end with some "Community Property" law that results in one party walking away with far more than he or she put into the relationship.  Every relationship carries risk.  This does not make it fault of the victim when that risk is realized.  Violence in a relationship is never appropriate, and it is never the fault of the victim.  Maybe she chose the relationship, but she did not choose the abuse, no matter how many time it happens.

The is a problem with rape as well.  Even judges, who should know better, frequently assume that the woman was "asking for it."  I can agree that she should have dressed more modestly, but that should have no bearing on the result of the case.  The rapist is the one who committed rape, not the victim.  The rapist is the one at fault for the crime, not the victim.  The victim could have been walking around town naked, and the crime would still be the rapists fault (though, indecent exposure laws do apply here, but they do not apply to the rape case).  How the woman dressed or acted has no bearing on whether rape is a crime or not.  If she did not consent, nothing else about her behavior and dress matters.

Victim blaming is a major problem in the U.S., especially with regards to domestic violence and rape.  Let me make it very plain what victim blaming is: Victim blaming is claiming that the behavior of the victim justified the crime.  Is dressing in skanky clothing really such an awful crime that it justifies the emotional harm and tearing away of personal freedom associated with rape?  Further, does wearing immodest clothing exempt a person from the legal right to representation, trial by a jury of peers, and freedom from cruel and unusual punishment?  Likewise, does making a error in judgment of the character of a person justify physical and emotional abuse, again, without legal representation and the rest of due process?  Even if it was the victims fault, vigilantism at this level is illegal, regardless of the supposed crime committed by the victim.  The fact is, in neither case has the victim broken the law, and in both, the abuser has.  Legally, it makes no difference whether the victim somehow "asked for it."  The behavior of the abuser was illegal.  Rape and domestic violence are both wrong, regardless of the situation.  Any person committing these crimes is a threat to society, and they should be treated as such.  When individuals or courts accept arguments about the victim "asking for it," they reinforce the bad behavior of the criminal.  This makes the abuser a bigger threat to society.  Victim blaming only makes the situation worse, and it punishes the person who deserves it the least.

Now, this also applies to economic abuse.  When I have discussed economic abuse with other conservatives, the most common reaction I get is, "Well, the employees agreed to those conditions, so there must be nothing wrong with it."  This is seriously wrong.  This is equivalent to saying that the abused woman agreed to the abuse by sticking around, so the abuser is not doing anything wrong.  Claiming that paying minimum wage is ethical, even though no one can survive on it, because the employees agreed to it is just another face of victim blaming.  This applies equally to those who are unemployed because they cannot find a job, or who are in poverty because they are not paid enough by their current job.  In case you are rich and have never had to find a job all on your own, giving up a job to look for another is not an option when you do not have any money, and many poor people work 40 hours a week at minimum wage (and statistically, they are working harder at their job than you have ever had to work in your entire life).  Most poor people do not have time to find a new job, when they already have one.

The point here is that being victimized is not the fault of the victim.  When another person chooses to act violently or unfairly, that person is the abuser, not the victim.  The fault of abuse belongs entirely to the abuser.  A victim that is forced or otherwise coerced into an abusive situation does not bear any fault for the abuse.  Our legal system is about protecting the innocent.  A rapist, a wife beater, and an unfair employer all harm the innocent.  We don't even bother protecting against the last one, but rapists and domestic abusers who are allowed to remain free frequently cause additional harm.  It does not matter how the victim was dressed, and in domestic abuse cases, it frequently does not even matter of the victim abandons the relationship.  Blaming the victim never helps protect society from additional harm.  Instead it makes the blamer look incompetent, it does further harm to the victim, and it reinforces the bad behavior in the abuser.

Small Businesses in the 3rd World

Recently, American investors have started funding startups in 3rd world countries.  It turns out that in Africa, the Middle East, and Southern Asia, there are a lot of artistic skills, without much local demand for the products.  Well, Americans have both the funds and the desire to purchase these products.  Some U.S. investors and charities have started providing funding for people in these 3rd world countries to start their own businesses based around these local arts.  Among other things, the funding is used to obtain training, pay for raw materials, and start shipping the goods to the U.S. where they can be sold.  Programs like this are helping to lift people in these countries out of poverty, and they are providing middle and upper class Americans with goods that are often of superior quality to factory produced products.  The problem is that this does not do anything for our large numbers of people in poverty here in the U.S.

Helping starving people in 3rd world countries is a noble goal.  Helping them in ways that reduce their dependence on us is even better.  It cannot last though.  We are not in any position to be giving people outside the U.S. significant amounts of our time and effort.  Poverty is increasing in the U.S., and we have plenty of our own starving people that need our help.  Our welfare system takes away some of the burden, but it adds others, partly because it is poorly designed.  What we really need is not U.S. investors providing funds for startups in Africa.  What we need is U.S. investors providing funding for U.S. poor.  Most of these businesses that we are funding in 3rd world countries could exist in the U.S. as well.

Americans can make shoes and rugs and many of the other things that we currently import.  Training is not that expensive, and it does not have to involve expensive trade schools.  Things like knitting and weaving can be learned on the internet.  Those with any level of artistic talent can learn how to make high quality costumes out of fairly cheap materials, and the current cosplay trend would help fetch some pretty good profits on this.  Things like blacksmithing and metal casting are also easy to learn on the internet, and in most areas of the U.S. there are groups that would be willing to help with training for cheap or even free.  This just brushes the surface.  There is a market in fixing old toys.  Learning to make simple web sites for individuals and small businesses is pretty easy.  Many people with unusual talents have managed to make a decent income just recording and publishing YouTube videos of their performances.  All of these things are artistic in nature, which means that they are not subject to the same sort of competition that normal jobs and products are.  There are only two things most poor Americans are missing: time and training (and sometimes equipment and materials).

American investors and charities should be looking at the American poor, instead of, or in addition to, those living in 3rd world countries.  They should offer funding, including enough for living expenses and training, for Americans who have or want skills in things that would provide the opportunity to start a small business based on artistic skills.  Besides helping those in the U.S., this could also be quite profitable for investors.  Demand for artistic products in the U.S. is fairly high, especially among the upper class.  Art knives made by reasonably skilled blacksmiths can demand $10,000 a piece or more, for products that take about one month to produce.  Hand crafted shoes can bring in a few hundred to a thousand dollars each for a week of work or less.  An old doll that takes a few hours to clean, repair, and repaint can bring in $20 or $30 with a half decent job (more for an experienced artist).  Hand knit or woven clothing and rugs can bring in a reasonable income as well.  Many people have managed to make a decent supplemental income with YouTube videos, and good ones can bring in ad revenue for months or years after they are published, without any additional work.  Even for those with less skill, these kinds of work can be done from home and supplement a poor income enough to make a significant difference.

What about those who are starving in 3rd world countries and do not even have food stamps to help them out?  The more we help our own poor, the more resources we will have to help others.  Every poor American we help out of poverty in a sustainable way is another potential donor, investor, or customer for poor people in other countries.  Instead of looking at investing in Americans as a drain from funds that could be invested elsewhere, consider it an opportunity for a more sustainable investment model for everyone in need.  Any investment that pays off will provide additional funding that can be used anywhere.  In addition, investors might encourage Americans they help to invest in or donate money to organizations that help people in other countries.  One rich person investing in some businesses in Africa will help a few people, until something happens that prevents that investor from continuing to invest (death or massive medical bills, for example).  Several thousand middle class Americans donating money to charities that help people start businesses in Africa will ultimately provide far better funding and be far more resilient.

The best strategy here is probably a mentorship strategy.  Investors should start investing in hobby-style artistic professions for poor Americans.  They should provide paid or volunteer mentors for every person they fund.  Those mentors should help beneficiaries manage their finances, find training, and setup their operations.  The mentors can also report back to the investors, to allow them to bail as soon as possible when a beneficiary is not using the funds wisely.  Another responsibility of a mentor should be to encourage beneficiaries to help out others once they are able to, and this theme should be repeated during regular meetings of the beneficiaries with the investors.  With mentors, chances of success will be much higher, and imminent failure will be much easier to detect much earlier.

The long term effect of this will be to teach Americans that cannot find jobs to create their own.  It will improve the U.S. economy, and it may even reduce the burden of the poor on the U.S. welfare system.  More importantly though, it will provide Americans with the means to provide aid for others far better than we are currently capable of.  A few rich investors helping the 3rd world poor start businesses to get them out of poverty cannot compete with what a robust U.S. economy composed of a large number of American middle class donors can do for them.  A common religious theme applies here: You must help yourself before you can truly help others.  To maximize the good we can do, we must first remove the mote from our own collective eye.