AI alignment - II of V

A group of researchers were building a model to better understand pneumonia. A hospital has to make one critical decision quickly - whether to treat the person as an inpatient or an outpatient. Pneumonia was at the time the sixth leading cause of death in the United States. So, correctly identifying which patients were at the greatest risk would result in a lot of lives being saved. The group had been given a dataset of about fifteen thousand pneumonia patients

One night, as a researcher was training the model, he noticed that it had learned a rule that seemed very strange. The rule was “If the patient has a history of asthma, then they are low-risk and you should treat them as an outpatient.” He didn’t know what to make of it because you don’t have to be a doctor to know that asthma is dangerous for a pneumonia patient. The doctors he consulted said, "We consider asthma such a serious risk factor for pneumonia patients that we not only put them right in the hospital . . . we probably put them right in the ICU and critical care." 

What was going on? The correlation that the system had learned was real. Asthmatics really were, on average, less likely to die from pneumonia than the general population. But the model had blindly noticed the correlation but didn’t know the reason - the positive correlation was precisely because of the elevated level of care they received. A researcher remarked, “So the very care that the asthmatics are receiving that is making them low-risk is what the model would deny from those patients." A model that was recommending outpatient status for asthmatics wasn’t just wrong; it was life-threateningly dangerous.

The researcher built another, more complicated model which seemed to work well but it too started giving strange results. It started saying that chest pain, heart disease and being over 100 is good for the patients when it obvious that they were not good for them. None of them made any more medical sense than asthma; the correlations were just as real, but again it was precisely the fact that these patients were prioritized for more intensive care that made them as likely to survive as the data showed. 

A department of the US government had sent data scientists to Afghanistan to analyze data -  financial records, movement records, cell phone logs, and more - to try to find patterns that would be useful to the war fighters. And they were already beginning to see that these machine-learning techniques were learning interesting patterns, but the users often didn’t get an explanation for why these patterns indicate something suspicious. 

Analysts had to put their names on the recommendation that goes forward. And they get scored based on whether that recommendation is correct. But they didn’t understand the rationale for the recommendation they were getting from the learning algorithm. Should they sign their name to it, or not? And on what basis, exactly, should they decide? As computing technology progresses, defense personnel have begun thinking about what risks and questions surround the idea of ever more autonomous weapons. 

As increasingly complex AI models keep getting deployed throughout the decision-making world, people have started recognizing how little they know about what’s actually going on inside those models. Whether it was getting rejected for a loan, being turned down for a credit card, being detained pending trial or denied parole, if a machine-learning system was behind it, you cannot be absolutely sure of how it arrived at the decision. 

In The Alignment Problem, Brian Christian gives the example of a Princeton cognitive scientist whose little daughter liked cleaning things. Once there were some chips on the floor, and she cleaned them up. He said to her, ‘Wow! Great job! Good cleaning! Well done!’ He thought that with the right praise, he would get some help in keeping the house clean. But it was not so simple. His daughter found the loophole in seconds. “She looked up at us and smiled,” he says, “and then dumped the chips out of the pan, back onto the floor, and cleaned them up again to try and get more praise.” This was a metaphor for how AI systems might do the wrong things with great speed and efficiency. 

The problem with machine learning systems was pointed out in 1960 by Norbert Wiener, a legendary professor at MIT and one of the leading mathematicians of the mid-twentieth century. In a paper, “Some Moral and Technical Consequences of Automation", here’s how he states the main point:

If we use, to achieve our purposes, a mechanical agency with whose operation we cannot interfere effectively . . . we had better be quite sure that the purpose put into the machine is the purpose which we really desire.

He further said, "It is my thesis that machines can and do transcend some of the limitations of their designers, and that in doing so they may be both effective and dangerous... Man and machine operate on two distinct time scales; the machine is much faster than man and the two do not gear together without serious difficulties." The computer does precisely what we tell it to do, just not what we thought we had told it to do. Much of software engineering is simply figuring out how to close the gap between those two things

AI alignment - I of V

AI is constantly in the news. More and more of the world is being turned over to various mathematical and computational models. Though they range widely in complexity, they are steadily replacing both human judgment and explicitly programmed software of the more traditional variety. Some do really cool things like discovering new molecules for medicine and some do really dark things like that AI meal planner app proposing a crowd-pleasing recipe featuring chlorine gas.

AI doesn’t mean one thing. There are chatbots, whose function is to output plausible looking text. You have image generators, whose function is to create images based on text input. Similarly for video generators. There are also systems designed to play games like chess or Go. There are systems designed to map from sequences of amino acids to predicted structures of the folded protein. There are systems that are designed to determine what goes into algorithmic feeds. 

When you open Google Maps, call Alexa or book an Uber you are dealing with a form of AI. The content on your social feeds or the ads that you that are targeted at you using AI. When you try to get a loan from a bank, you are screened by AI. What price you pay for your home, or your car insurance, are decided by AI. When you are interviewing for a job, your face and responses may be analysed by AI. 

What all of these things do have in common is that they are the result of doing statistical processing over large data sets. But the input data that's used to create the systems are different. The kind of statistical patterns that are being mapped are different. Just saying "AI" gives the impression that there's one thing out there and it knows "about the shape of folded proteins", and also about "how to play chess", and it knows the answer to whatever question you might put into the chatbot. That makes it seem like it's one super intelligent entity when it's actually a bunch of separate software programs designed by different people, trained on different data for different purposes. 

There was a time when most artificial intelligence was programmed by computer scientists. And then scientists figured out how to get AI to learn how to do what we instructed it to do but we still would provide them with the instructions that define the goal of the AI model. In other words, they got a digital computer to improve of its own accord. By developing machines that could learn by human instruction or their own experience, they removed the need for programming.

This gave rise to a new issue, the alignment problem viz. whether the AI is reaching its intended goal or giving some unintended result. In the last five years or so, these fears have started coming to life. We are living in a world full of examples of this - image recognition software that captioned a selfie of two black Americans as "gorillas", or  self-driving cars that fail to identify jaywalking pedestrians and end up causing fatal collisions. Broadly, we can think of a machine learning system as having two halves. Each of these halves offers an opportunity for things to become misaligned: 

1. There is the training data, the set of examples from which the system learns. The AI is then at the mercy of the examples from which it is taught. If a certain type of data is underrepresented or absent from the training data but present in the real world, then things will go wrong. 
2. The objective function, which is how we are going to mathematically define success in each of those examples. It basically tells them what we want it to do. 

Take the 2018 crash of the Uber car that killed a pedestrian in Arizona. The system was built on an object classification system that had a very rigid set of categories that included pedestrian, cyclist, debris, etc. and had thousands of examples of each of those things. The system did not have any training data of jaywalkers so it was unprepared to encounter someone crossing a road not at a crosswalk. But this particular woman was walking a bicycle across the street, which was something that the system had never seen causing a fatal crash. The model is only as good as what data was put into it.

Ethics and Modern gene therapy - IV of IV

If CRISPR became a standard tool in fertility clinics, people might lose their suspicions of it — just as people lost their suspicions of in vitro fertilization in the 1980s. Before long, people might be willing to entertain a new use for CRISPR. Doctors might edit beneficial changes into an embryo’s genes. Parents could give their children all the advantages that scientists have found in our species’ genetic variations. 

Since there are always advances in science, parents might postpone having children in the hope that new variations may be found which will give their children better advantages. This will make decisions about when to have children seem the same way as how people wait to buy a phone until a new model is released. The ethicist Robert Sparrow argues that this might lead to a sense of genetic inferiority for earlier generations. He wonders if future generations might find themselves stuck in an “enhanced rat race.”

As is always the case, the problem is the system. If success depends on intelligence, and intelligence can be engineered, then parents feel morally compelled to enhance their children. Parents genetically enhance their children out of love but that love becomes entangled with fear and competition. Some children will suffer or die but they will reason that it is the price of staying competitive. Merit stops being “fair” and becomes biologically rigged from the start. Ethical boundaries shift easily when success is at stake. The most dangerous futures aren’t imposed — they’re gradually accepted. Over time, what once seemed extreme becomes “just how things are.”

This might lead to unfamiliar legal territory. A few cases have been brought by children in the US against their parents for allowing them to be born with congenital diseases. According to these “wrongful life” lawsuits, the parents were negligent for ignoring tests on the fetus before birth and going ahead with it anyway. Some ethicists now wonder if children in the future may sue their parents for not using the latest genetic engineering engineering techniques thereby putting them at a disadvantage with respect to future generations 

In The Case Against Perfection, Michael Sandel, an American political philosopher, argues against enhancement. He says that if bioengineering made the myth of the "self-made man" come true, it would be difficult to view our talents as gifts for which we are indebted, rather than as achievements for which we are responsible. What would be lost if biotechnology dissolved our sense of giftedness? This would make us less likely to view our traits as a matter of chance. He writes: 

A lively sense of the contingency of our gifts — a consciousness that none of us is wholly responsible for his or her success - saves a meritocratic society from sliding into the smug assumption that the rich are rich because they are more deserving than the poor. Without this, the successful would become even more likely than they are now to view themselves as self-made and self-sufficient, and hence wholly responsible for their success. Those at the bottom of society would be viewed not as disadvantaged, and thus worthy of a measure of compensation, but as simply unfit, and thus worthy of eugenic repair. The meritocracy, less chastened by chance, would become harder, less forgiving.

He gives an example of the real world consequences. Consider insurance. Since people do not know how their fate will pan out, they pool their risk by buying health insurance and life insurance. The actual result is that, over time, the healthy wind up subsidizing the unhealthy, and those who live to a ripe old age wind up subsidizing the families of those who die early. What ends up happening is that that people pool their risks and resources and share one another's fate.

But insurance markets work properly only as long as people do not know or control their own risk factors. Suppose genetic testing advanced to the point where it could reliably predict each person's medical future and life expectancy. Those confident of good health and long life would opt out of the pool, causing other people's premiums to skyrocket. The insurance market will collapse as perfect generic knowledge ends up separating those with good genes from the company of those with bad ones.

One important ethical issue is that the use of such technologies will support ongoing inequalities among military parties. CRISPR is currently an expensive technology. Some developed countries might think of using this technology to further strengthen their defenses and even attack underdeveloped or developing countries. The US military started a program called Safe Genes to gene modify organisms to be used in battle and anti-CRISPR tools to disable bio-weapons. This situation could cause a constant tension, making it difficult to provide an environment of peace and stability worldwide. 

There is yet another aspect of the genetic editing of microorganisms to consider, as CRISPR could also be used to synthesize and manipulate pathogens, including smallpox, the Spanish flu virus, avian H5N1 flu virus, and SARS. Anyone with the appropriate equipment could engineer a vaccine-resistant flu virus or invasive species in a crude laboratory. Bio-terrorists might use it to turn common microbes into a pathogenic weapon.

I heard of an economics professor who was teaching macroeconomics (I think it was  Gregory Mankiw). He told the students (quoting from memory), ‘Both of us are confused. The only difference is that you are naively confused and I am profoundly confused.’ After this brief discussion about CRSPER ethics, I hope you are profoundly confused.

"May you live in interesting times" is an English expression that is claimed to be a translation of a traditional Chinese curse. The expression is ironic: "interesting" times are usually times of trouble. With climate change, AI, and CRISPR, 2050 promises to be very interesting indeed, perhaps more interesting than anyone had bargained for. (2050 seems to be too far in the future but it is a nice number!)

Ethics and Modern gene therapy - III of IV

There are some genes that have both positive and negative effects in different contexts. For example, researchers now suspect that people who carry one copy of the mutated gene that causes cystic fibrosis (which requires two copies) have an increased defense against tuberculosis. Even gene variants implicated in neurodegenerative diseases like Alzheimer’s may have benefits, such as improved cognitive function and better working memory in young adults. What decisions would you make? 

Schizophrenia, depression, and bipolar disorder can be brutal, often deadly. While trying to eliminate similar disorders, we should consider whether there might be some cost to society, even to civilization. A reason that scientists will not eliminate conditions such as psychiatric disorders or conditions such as autism is that some of the risk for these disorders almost certainly comes in trade for small competitive advantages, such as heightened sensitivity, concentration, or openness to experience.

A study showed a 77 percent rate of psychiatric disorders in eminent fiction writers. Writers are 10 times, and poets 40 times, more likely to be bipolar than the general population. Vincent van Gogh had either schizophrenia or bipolar disorder. So did the mathematician John Nash. People with bipolar disorder include Ernest Hemingway, Mariah Carey, Francis Ford Coppola, Graham Greene, Sylvia Plath, Edgar Allan Poe, and hundreds of other artists and creators. 

To what extent does dealing with mood swings, fantasies, delusions, compulsions, mania, and deep depression help spur, in some people, creativity and artistry? Would you cure your own child from being schizophrenic if you knew that, if you didn’t, he would become a Vincent van Gogh? We have to face the potential conflict between what is desired by the individual versus what is good for human civilization. 

A reduction in mood disorders would be seen as a benefit when seen from the point of view of an individual and as a cost when seen from the point of view of society. As we learn to treat mood disorders with drugs and eventually with genetic editing, will we have more happiness but fewer Hemingways? Do we wish to live in a world in which there are no Van Goghs? But what moral right do we have to require another family to forgo a desired genetic intervention simply for the sake of adding to the diversity of society? 

Decisions about genetic editing are likely to be driven by consumer choice and the persuasive power of marketing. Initially people will think that if we can do so safely, why shouldn’t we prevent abnormalities, diseases, and disabilities? That sounds reasonable and morally justified but it might prove to be a slippery slope. They will naturally start thinking: Why not improve our capabilities and create enhancements - changes in which DNA is altered not to correct a harmful gene variant but to provide some type of genetic advantage, perhaps high intelligence or athletic abilities. (Of course, there is a limit to what enhancements will be possible or safe to attempt.)

While thinking about correcting disabilities, we should keep one factor in mind: to what extent they are inherently disabling and to what extent the disadvantage is due to our social constructs and prejudices. The disadvantages from being deaf, for a human or any other animal, are very real. In contrast, any disadvantages to being gay or Black are due to social attitudes that can and should be changed. That is why we can make a moral distinction between using genetic techniques to prevent deafness and using these techniques to influence such things as skin color and sexual orientation.

Then comes the question of super-enhancements.  These are traits and capacities that exceed what any human has ever had.  Suppose people can choose for their kids to have super-eyesight? What about adding the capacity to see infrared light or some new color? DARPA, the Pentagon’s research agency, already has a project going to study how to create genetically enhanced soldiers.

For example, genetic enhancement may be possible for improving memory. Scientists have managed to manipulate a memory-linked gene in fruit flies. They have produced smart mice by inserting extra copies of a memory-related gene into mouse embryos and the improvement was passed on to offspring. Human memory is more complicated. Should research in this area be allowed? But the natural instinct of scientists is to pioneer procedures and make discoveries. If a nation imposes too many. restrictions, its scientists will move elsewhere and pursue the research. 

Since the wealthy would be able to afford the procedure more often, and since any beneficial genetic modifications made to an embryo would be transmitted to all of that person’s offspring, linkages between class and genetics would keep growing from one generation to the next, no matter how small the disparity in access might be. Consider the effect this could have on the socioeconomic fabric of society. The co-discoverer of CRISPR, Jennifer Doudna says, 

We could create a gene gap that would get wider with each new generation...If you think we face inequalities now, imagine what it would be like if society became genetically tiered along economic lines and we transcribed our financial inequality into our genetic code.

This may also create a different kind of injustice. Using gene editing to “fix” things like deafness or obesity could create a less inclusive society, one that pressures everyone to be the same. Part of what makes our species unique, and our society so strong, is its diversity. A fear is that gene editing will increase existing prejudices against people who fall outside a narrow range of genetic norms. 

Ethics and Modern gene therapy - II of IV

There can be various technical difficulties in producing designer babies. Thousands of genetic variations can influence complex traits, psychiatric risk, personality traits, and capacities such as human intelligence. Take any given genetic variant. None has more than a fraction of a single percentage point of an effect on the risk for a psychiatric disorder or condition. 

Each of the variants in our genes can have enhancing or diminishing effects on other genes depending on the context in which they are inherited. Genetic variants may be deleterious in some cell types, such as neurons, but advantageous in other cell types, such as immune cells. A lot of scientific evidence shows that chronic stress and poverty contribute to alterations in brain circuitry and blood pressure, dramatically influencing health and mortality.

A gene often has three or four different functions, so altering a single gene may have three or four effects. A gene that builds a protein named “protein S” is a blood coagulant, but it was recently shown to have a critical role in regulation of the immune system. The opposite is also true: multiple biological codes or parts can perform the same function. To engineer new systems would require a complete analysis of an entire network, not just a single gene. 

For argument’s sake, let us assume that all these difficulties will be overcome. And we are not talking of the distant future. The time frames being talked about are 15-20 years. If so, what sorts of ethical issues will humanity have to face? In The Code Breaker, Walter Isaacson discusses some thought experiments, which give a flavor of the kinds of questions that we may have to grapple with. 

Sharon Duchesneau and Candy McCullough wanted a sperm donor so they could conceive a kid. That sounds straightforward until you are told that both of them are deaf and lesbians and they wanted a child who is also deaf. They consider their deafness to be part of who they are rather than something to be cured, and they wanted a child who would be part of their cultural identity. So they advertised for a sperm donor who was congenitally deaf. They found one, and now they have a deaf child.

Some people condemned them for making a child disabled intentionally but the deaf community appreciated their action. Where do you stand on this? Should they be praised for preserving a subculture that contributes to the diversity? Would it have been ok if, instead of using a deaf sperm donor, the couple had used pre-implantation diagnosis to select an embryo that had the genetic mutation for deafness? What if they had safely destroyed the child’s eardrums after birth?

Now let us look at gene editing that is done to enhance the traits of our children. The MSTN gene produces a protein that reduces muscle growth when they reach a normal level. Suppress the gene and muscle growth is in overdrive. This has already been done to produce “mighty mice" and cattle with “double muscling". Pushy parents and athletic directors who want champion athletes would be very interested. By performing germline editing, they might produce athletes with bigger bones and stronger muscles. 

When athletes cheat by using steroids, we find it easy to say that they should be banned. But what do we do if athletes' prowess comes from genes they were born with? For example, almost every champion runner has what is known as the R allele of the ACTN3 gene. It produces a protein that builds fast-twitch muscle fibers, and it is also associated with improving strength and recovery from muscle injury.

Someday it may be possible to edit this variation of the ACTN3 gene into the DNA of your kids. Would that be unfair? Does it matter if those genes were paid for by their parents rather than bestowed by a random natural lottery? In future, would we end up admiring the wizardry of the genetic engineers of athletes rather than the diligence of the athletes?

Ethics and Modern gene therapy - I of IV

In Kazuo Ishiguro's novel, Klara and the Sun, the “lifted” are children who have undergone a genetic enhancement procedure designed to increase intelligence and academic ability. It’s something wealthier families choose for their children to secure better futures — elite education, careers, and status. Most top universities in the novel’s world primarily accept lifted students, creating a strong incentive to undergo the process. The parents take this risk in spite of the possibility of the procedure causing illness or even death. Such a dystopian world may not be as far in the future as you might think.  

Genetic engineering has been practiced for five decades. It is the process of altering an organism's genome to change its characteristics in a particular way. It has been used to make food more nutritious, create synthetic insulin and provide promising treatments for illnesses including leukemia and sickle cell disease. Modern gene therapy is being used to treat eye diseases which can cause blindness, promote the growth of healthy skin or add supplementary copies of working genes that fix rare blood or immune system disorders.

Enter CRISPR. Remember the name. I am sure you are dying to know what it stands for so here it is: Clustered Regularly Interspaced Short Palindromic Repeats. CRISPR makes editing genomes much more precise, cheap, and easy than was possible earlier. The technique is considered so significant that the discoverers, Jennifer Doudna and Emmanuelle Charpentier, won the Nobel Prize in Chemistry in 2020, less than a decade after the discovery, something that rarely happens. Biologists began speaking about their life before and after CRISPR.

CRISPR is sold on the internet in kits, and is actively being used to do trivial things, such as to create fluorescent beer. Its ease of use has also raised concerns about “biohackers” who view gene modification as a right and alter microbes and organisms. “Mail-Order Crispr Kits Allow Absolutely Anyone to Hack DNA,” declared the headline of a November 2017 article in Scientific American. The iconoclast scientist Josiah Zayner has used CRISPR to hack into his own genes. (There is a docuseries on Netflix called "Unnatural Selection" where you can see it.)

There are even CRISPR jokes: Why has KFC asked scientists to edit the chicken genome? Because they want something CRISPR. And who is CRISPR's favorite actor? Gene Hackman

So what is the fuss all about? For that, first a little bit of biology. The body contains two types of cells: somatic and germ line cells. Somatic cells refer to any cell of a living organism other than the reproductive cells. The reproductive cells - the egg and the sperm - are called the germ line cells. A germ line cell passes on to the next generation while somatic cells don’t. 

CRISPR is so precise that gene therapy in people with devastating illnesses seems feasible. For example, physicians could directly correct a faulty gene, say, in the blood cells of a patient with sickle-cell anemia. But that kind of gene therapy wouldn’t affect germ cells, and the changes in the DNA wouldn’t get passed to future generations.

In contrast, the genetic changes created by germ-line engineering would be passed on, and that’s what has made the idea seem so objectionable. “Germ line” is biologists’ jargon for the egg and sperm, which combine to form an embryo. By editing the DNA of these cells, it could be possible to correct disease genes and pass those genetic fixes on to future generations. Such a technology could be used to rid families of scourges like cystic fibrosis. 

Germline genome editing leads to many bioethical issues. For example, what to do if the editing leads to occurrence of undesirable changes in the genome? Can parents give informed consent for editing the genomes of unborn children? If not, from whom do you obtain the consent? The counterargument is that parents already make many decisions that affect their future children, including similarly complicated decisions with IVF. Another fear is that germ-line engineering is a path toward a dystopia of superpeople and designer babies for those who can afford it. Want a child with blue eyes and blond hair? Why not design a highly intelligent group of people who could be tomorrow’s leaders and scientists?

Others believe the idea is dubious because it’s not medically necessary. It’s already possible to test the DNA of IVF embryos and pick healthy ones, a process that adds about $4,000 to the cost of a fertility procedure. A man with Huntington’s, for instance, could have his sperm used to fertilize a dozen of his partner’s eggs. Half those embryos would not have the Huntington’s gene, and those could be used to begin a pregnancy.

George Church, a geneticist at Harvard, likes to show a slide on which he lists naturally occurring variants of around 10 genes that, when people are born with them, confer extraordinary qualities or resistance to disease. One makes your bones so hard they’ll break a surgical drill. Another drastically cuts the risk of heart attacks. Church proceeded to tell the audience that he thought changing genes “is going to get to the point where it’s like you are doing the equivalent of cosmetic surgery.”

Regulations about gremline editing are variable and often lack teeth. For example, in many countries like Canada, France, Germany, Brazil, and Australia, clinical interventions in the human germline are expressly prohibited, with criminal sanctions that range from fines to lengthy prison terms. In other countries, such as China, India, and Japan, these interventions are forbidden, but with guidelines that are less enforceable. In the United States, there are no outright bans but any clinical trials would need to receive regulatory approval by the Food and Drug Administration.

There’s a risk that overly restrictive policies in some countries will encourage what might be called CRISPR tourism in others. Patients with means could travel overseas to jurisdictions where regulations are more forgiving or absent altogether. Excessive restrictions on research might lead scientists to continue their experiments behind closed doors. Trying to find a balance between maintaining regulatory environments that permit research and clinical applications but strict enough to prevent the worst excesses would be tough. 

Hypocrisy is not all bad

There is an increasing tendency to accept shocking statements by politicians by just saying that they are being authentic and not hiding behind hypocritical statements. Political leaders used to at least pretend that they are doing the right things some of the time. But Trump has been so successful in making people familiar with the idea of not pretending that they now just shrug their shoulders and say that Trump is being Trump. 

The global system shaped after World War II was built around open markets, human rights, international institutions like United Nations and cooperation and rule-based norms. A large part of the world did not accept it. There were many situations when the system was ignored more than being followed, particularly by the United States itself. But you still had this as the kind of default operating system of the international world.

Whenever the United States did not live up to those principles, it always tried to frame its actions as if it was trying to uphold them. So for example, for the war in Iraq, the Bush administration went to the United Nations, tried to get resolutions, had inspectors put in place, gathered a coalition of 40 plus nations, went to the United States Congress, and then went to war with Iraq. The war may have been misguided, but there was an effort to put it in the context of this larger international order that the United States believed in and was part of.

Now it has gone from being a country that believed in the international system that it had put into place to one that openly violates it. "Openly violates it" is the part that is important. For the current war in Iran, there was no effort to go to the United Nations or to go to Congress. The United States has exactly one ally, Israel. This was deliberate. The Trump administration doesn't believe in any of those features. It wants the unilateral exercise of American power for American national interests as it conceives it to be.

The practice of filling the government with incompetent loyalists has been going on for thousands of years and people know that it will always be there. But some excuse to show that you're doing it for other reasons will generally be given to cover up the actual reason for doing it. But now even this pretense is often not required. Is this a good thing? 

The Oxford English Dictionary tells us that hypocrisy is the “practice of claiming to have moral standards or beliefs to which one’s own behavior does not conform; pretense.”  It is generally viewed as a negative trait; a significant moral failing, especially in a leader. It is often seen as a mark of dishonesty and a lack of authenticity. But it easy to miss the good about hypocrisy - even giving lip service to an ideal that you fall short of maintains the idea that the ideal should remain and people should aspire for it. 

If people were required to perfectly live up to ideals of honesty and compassion at all times for those ideals to exist, there would be no ideals at all. According to Gandhi, there must always be an unbridgeable gulf between the ideal and its practice. The ideal will cease to be one if it becomes possible to realise it. He argues: "Where would there be room for that constant striving, that ceaseless quest after the ideal . . . if mortals could reach the perfect state while still in the body?"

The maxim that 'hypocrisy is the tribute that vice plays to virtue' makes the same point - you're only truly capable of hypocrisy if you're to some degree accepting the importance of certain norms. It's by reference to those norms that you can be called a hypocrite. Hypocrites who fail to keep their promises but refuse to abandon the ideals they betray help keep those standards in place for society to strive toward. The social condemnation of hypocrisy reinforces moral norms and promotes more authentic and accountable behavior in society.

Some situations may require hypocritical behavior in order to reduce tensions in social relations. When citizens appear to conform to the social and cultural conventions and norms of their communities, where their instincts and desires are repressed, they cannot merely be accused of being hypocritical.  Living in a group may require compromise at certain times. When politicians appear hypocritical, they may be performing much better than if they remained steadfast in their consistent adherence to principles. For example, when the leaders of various countries praise Trump to the skies, you know that they are lying but you also know that it is the best way to get a good deal for their countries. 

When a person is accused of hypocrisy, it makes both the charging party and those being charged critically reflect on the action. Trump-style dismissal of any appeal to ethics and virtues, or the belief that such an appeal is inherently in bad faith breeds cynicism and a decline in social standards.  A cynical agreement in society that hypocrisy is a common occurrence and that we are all hypocrites some of the time reduces the effective functioning of a society. 

Anne Applebaum writes that some countries are members of what she calls Autocracy, Inc. - Myanmar, Zimbabwe, Iran,  Cuba, Venezuela, China, Russia etc. They have spent many years disputing the human rights language long used by international institutions. They dismiss treaties and conventions on war and genocide, and concepts such as “civil liberties” and “the rule of law” as embodying Western ideas that don’t apply to them. They feel no shame about the use of open brutality and send hundreds of their citizens to their deaths.

Once upon a time, the leaders of the Soviet Union, the most powerful autocracy in the second half of the twentieth century, cared deeply about how they were perceived around the world. They vigorously promoted the superiority of their political system, and they objected when it was criticized. They at least paid lip service to the aspirational system of norms and treaties set up after World War II, with its language about universal human rights, the laws of war, and the rule of law more generally. Even in the early part of this century, most dictatorships hid their true intentions “behind elaborate, carefully manipulated performances of democracy". But all that pretense is now not required. 

The Overton Window is a model for understanding how ideas in society change over time and influence politics. It was developed in the 1990s by Joseph Overton, a political scientist. The window illustrates the general public’s most acceptable policies in the center and the more untenable policies on the ends. According to the concept, politicians are limited in what policy ideas they can support — they generally only pursue policies that are widely accepted throughout society as legitimate policy options. These policies lie inside the Overton Window. 

Politicians and others in the political arena might shift or expand the span of the Overton window to make specific policies more or less acceptable in public opinion. Politicians of various countries, by their statements and actions over a number of years, have shifted the Overton window towards reduced importance of a number of moral ideas. Anne Applebaum writes in Autocracy Inc.:

This is the core of the problem: the leaders of Autocracy, Inc., know that the language of transparency, accountability, justice, and democracy will always appeal to some of their own citizens. To stay in power they must undermine those ideas, wherever they are found.

Russia and China would not have dreamt that they would have a person in the White House who would do their job for them. They will be content to follow a famous strategic maxim attributed to Napoleon Bonaparte - "Never interrupt your enemy when he is making a mistake".