AI Quick Hits

1) I really enjoyed this recent post from AI guru Ethan Mollick, “On-boarding your AI Intern”

In previous posts, I have made the argument that, for a variety of reasons, it is better to think of AI as a person (even though it isn’t) than a piece of software. In fact, perhaps one of the most interesting aspects of our current AI moment is that several billion people just got free interns. They are weird, somewhat alien interns that work infinitely fast and sometimes lie to make you happy, but interns nonetheless.

So, how can you figure out how to best use your intern

1

? Just like any new worker, you are going to have to learn its strengths and weaknesses; you are going to have to learn to train and work with it; and you are going to have to get a sense of where it is useful and where it is just annoying. The stakes for this are quite high. People using AI have 30-80% higher productivity in some writing and coding tasks, and often feel happier having offloaded their most annoying work. That is a big incentive to learn to work with your intern.

What would an AI intern be great for?  Choosing the best excerpts of articles for quick hits. So, let’s see how it goes.  If you don’t like the excerpts, you know who to blame.

2) It’s crazy how Scientific American is far more interested in pushing an ideological agenda than interesting science these days.  The latest was sharing this fascinating article about White-throated sparrows as somehow relevant for human gender debates (it’s really interesting on its own):

The White-throated Sparrow is common and familiar, hopping on the ground under bird feeders all over the eastern states in winter. But this seemingly ordinary backyard bird has a secret identity—or, actually, four secret identities. And it’s these multiple personalites that place the White-throat at the center of mysteries scientists are still working out.

Watch a flock of White-throats in spring and you’ll notice they have two kinds of head patterns. Some wear snappy stripes of black and white across the top of the head. Others have more modest head stripes of dark brown and tan. That superficial difference might not seem like a big deal, but it reflects a remarkable divergence in the lifestyles of these individuals.

As Lowther discovered, mated pairs of White-throats almost always involved one bird of each color morph: Either a tan-striped male with a white-striped female, or a white-striped male with a tan-striped female. Intrigued, Lowther extended his research, joined by biologist J. Bruce Falls and others.

They found that the color differences were more than skin deep. The two morphs had different personalities, different behaviors, different hormones, and even different chromosomes.

3) On the 25th anniversary of Seinfeld:

But they also presented an irreverent version of adulthood that I had never seen on TV or in life: a playful yet sophisticated world where grown-ups joked and laughed together and didn’t take themselves too seriously, even when everyone around them was being very serious indeed.

For the somehow uninitiated, “Seinfeld,” created by Jerry Seinfeld and Larry David, stars Seinfeld as a fictionalized version of himself and follows his shenanigans with his three closest friends: his childhood buddy, George Costanza (Jason Alexander); his former girlfriend turned pal, Elaine Benes (Julia Louis-Dreyfus); and his oddball neighbor, Kramer (Michael Richards). It is regarded as one of the greatest shows of all time.

It has consistently been framed as a comedy about four terrible people, with good reason. Jerry and his fellow misfits lied, cheated and stole. They were petty and shallow. They created a framework for “bad” sitcom characters that shows like “It’s Always Sunny in Philadelphia” would embrace with great relish and success.

But what if they were also onto something? What if their refusal to conform to the expectations of adulthood — marriage, children, career advancement — was not just a sign of immaturity, but also a form of resistance? What if their rejection of the conventional markers of success was not just a flaw, but also a strength?

3) Not going to have GPT summarize an abstract, though, “Individual Empowerment, Institutional Confidence, and Vaccination Rates in Cross-National Perspective, 1995 to 2018”

In the past decade, before the onset of the Covid-19 pandemic, rates of childhood vaccination against diseases such as measles, diphtheria, pertussis, and tetanus declined worldwide. An extensive literature examines the correlates and motives of vaccine hesitancy—the reluctance or refusal to vaccinate despite the availability of vaccines—among individuals, but little macrosociological theory or research seeks to explain changes in country-level vaccine uptake in global and comparative perspective. Drawing on existing research on vaccine hesitancy and recent developments in world society theory, we link cross-national variation in vaccination rates to two global cultural processes: the dramatic empowerment of individuals and declining confidence in liberal institutions. Both processes, we argue, emerged endogenously in liberal world culture, instigated by the neoliberal turn of the 1980s and 1990s. Fixed- and random-effects panel regression analyses of data for 80 countries between 1995 and 2018 support our claim that individualism and lack of institutional confidence contributed to the global decline in vaccination rates. We also find that individualism is itself partly responsible for declining institutional confidence. Our framework of world-cultural change might be extended to help make sense of recent post-liberal challenges in other domains.

4) Good stuff from NYT, “The Greatest Wealth Transfer in History Is Here, With Familiar (Rich) Winners”

n 1989, total family wealth in the United States was about $38 trillion, adjusted for inflation. By 2022, that wealth had more than tripled, reaching $140 trillion. Of the $84 trillion projected to be passed down from older Americans to millennial and Gen X heirs through 2045, $16 trillion will be transferred within the next decade.

…

The pandemic has only accelerated this trend. The stock market has soared to record highs, while home prices have risen at their fastest pace in 15 years. These gains have disproportionately benefited older Americans who own more stocks and real estate than younger generations.

…

The result is a widening gap between the haves and have-nots that is likely to persist as wealth is handed down from one generation to the next. According to a study by the Federal Reserve Bank of St. Louis, the wealthiest 10 percent of American families owned 77 percent of total family wealth in 2019, up from 71 percent in 1989. The bottom half of families owned just 2 percent of total wealth, down from 4 percent in 1989.

…

The concentration of wealth among a few families also raises concerns about the influence of money on politics and democracy. Some of the richest heirs in America, such as Charles Koch and George Soros, have used their fortunes to fund political causes and candidates that align with their views.

5) The kids and their subtitles these days!

Recent research is showing that the use of subtitles on TV has continued to grow, with people choosing to use them. Why is this? If the speech intelligibility of the content we mix is so bad, surely we cannot be doing our job properly. What is going wrong?

…

The BBC has been conducting research into this issue and has found that subtitle usage has increased from 7.5% in 2007 to 18% in 2016. However, this figure does not include online viewing, where subtitle usage is much higher. According to Netflix, more than 80% of its UK users watch with subtitles on.

…

The BBC research also found that the main reasons for using subtitles were not related to hearing impairment, but rather to factors such as background noise, accents, mumbling and fast speech. Some viewers also said they used subtitles to help them understand complex plots or unfamiliar vocabulary.

6) One more “Jury Duty” episode to go for me.  So good!

Jury Duty—a series starring mostly unknown performers, tucked away on a largely unknown streamer—is incredible reality television, a boundary-pushing hidden-camera program. Set inside a fake courtroom, the show follows Ronald, a guy who believes he’s participating in a documentary about jury duty but who is actually surrounded by actors roping him into progressively weirder scenarios.

…

Jury Duty has become a word-of-mouth hit, and Ronald a bona fide star. According to a JustWatch report, the show was the most popular streaming series the week of its finale in April, nabbing more viewers than Netflix’s Beef and The Diplomat. Ronald, meanwhile, just appeared in an ad with Ryan Reynolds.

…

Given the show’s triumphs, the producers have teased the possibility of a second season; they told Variety that the best aspects of their concept are “infinitely repeatable.” But as true as that may be—other hoax-driven series in the past, such as Spike’s The Joe Schmo Show, ran for multiple seasons—creating more Jury Duty would be a shame.

…

The magic of Jury Duty is that it doesn’t yet have a formula. It’s an experiment that worked because of its novelty and unpredictability. To repeat it would be to risk losing what made it so special in the first place.

7) Another abstract from some really interesting PS research, “Who Supports Political Violence?”

The last few years have witnessed an increase in democratic “backsliding” in the United States—a decline in the quality of democracy, typically accompanied by an influx of non-normative behavior, such as political violence. Despite the real consequences of support for violence, fairly little is known about such an extremist attitude outside studies of terrorism or aggression. Using a unique survey containing many psychological, political, and social characteristics, we find that perceived victimhood, authoritarianism, populism, and white identity are the most powerful predictors of support for violence, though military service, conspiratorial thinking, anxiety, and feelings of powerlessness are also related. These patterns suggest that subjective feelings about being unjustly victimized—irrespective of the truth of the matter—and the psychological baggage that accompanies such feelings lie at the heart of support for violence. We use these results to build a profile of characteristics that explain support for violence; the predictive validity of this profile is then tested by examining its relationship with support for the January 6, 2021, U.S. Capitol riot, with which it is strongly associated, even accounting for support for Donald Trump. Our findings have implications for the detection of extremist attitudes and our understanding of the non-partisan/ideological foundations of anti-social political behavior.

8) Noah Smith, “How technology has changed the world since I was young”

he world has changed a lot since I was young. Technology has changed it. And I’m not just talking about the internet and smartphones and social media. I’m talking about the deeper changes that have reshaped our society and our culture, our economy and our politics, our values and our beliefs.

…

The first big change is that technology has made us more connected than ever before. We can communicate with anyone, anywhere, anytime, with a click of a button or a swipe of a screen. We can access a vast amount of information and entertainment, from news and podcasts to movies and games. We can share our thoughts and feelings, our opinions and experiences, our likes and dislikes, with millions of strangers online.

…

The second big change is that technology has made us more powerful than ever before. We can create and manipulate things that were once beyond our imagination, from artificial intelligence and biotechnology to nanotechnology and quantum computing. We can solve problems that were once unsolvable, from curing diseases and exploring space to fighting climate change and enhancing human capabilities. We can influence and shape the world around us, for better or for worse.

…

The third big change is that technology has made us more uncertain than ever before. We face new challenges and risks that we don’t fully understand or control, from cyberattacks and misinformation to ethical dilemmas and social unrest. We face new questions and choices that we don’t have clear answers or guidelines for, from privacy and security to identity and morality. We face new possibilities and scenarios that we don’t have adequate preparation or foresight for, from technological singularity and superintelligence to posthumanism and transhumanism.

9) Don’t know how I missed this from 2021, but it’s excellent, “Reducing gun violence: What do the experts think?”

Gun violence is a complex and multifaceted problem that requires a comprehensive and evidence-based approach. Unfortunately, the public debate on this issue is often polarized and simplistic, pitting gun rights against gun control, or law enforcement against community prevention. This binary framing obscures the diversity of perspectives and experiences among those who are most affected by gun violence, as well as the potential for common ground and collaboration among stakeholders.

…

To move beyond this impasse, we convened a group of experts from different disciplines and backgrounds to discuss what we know and don’t know about reducing gun violence, and what policies and programs are most promising and feasible. The group included researchers, practitioners, advocates, and policymakers who have worked on various aspects of gun violence prevention, such as public health, criminal justice, mental health, education, and civil rights.

…

The group agreed on several key points:

• Gun violence is not a monolithic phenomenon, but rather a collection of different types of violence that vary by context, motive, means, and impact. Therefore, no single policy or program can address all forms of gun violence; instead, we need a portfolio of interventions that are tailored to specific populations and settings.

• Gun violence is not only a criminal justice problem, but also a public health and social justice problem. Reducing gun violence requires addressing its root causes and risk factors, such as poverty, inequality, trauma, racism, and social isolation.

• Gun violence prevention should not be driven by ideology or politics, but by data and evidence. We need more research to identify the most effective and cost-effective strategies to prevent gun violence, as well as to evaluate the implementation and impact of existing policies and programs.

10) And a great post from Yglesias on policing:

The basic problem with policing in America is that it’s not very effective at preventing crime. The clearance rate for homicides is only about 60%, and for other violent crimes it’s much lower. That means that most criminals get away with their crimes, and most victims don’t get justice.

…

One reason for this low effectiveness is that police officers are not allocated to the places where they are most needed. In a new paper, Tanaya Devi and Roland Fryer show that there is a large spatial mismatch between where police officers are deployed and where crime occurs. They use data from 242 U.S. cities to measure the number of officers per square mile in each census block group, and compare it to the number of crimes per square mile in the same area.

…

They find that there is a negative correlation between police presence and crime: Areas with more crime have fewer officers per square mile, and vice versa. This correlation is especially strong for violent crimes like homicide, robbery, and aggravated assault. They estimate that reallocating officers to match the spatial distribution of crime could reduce homicides by 11% and violent crimes by 7%, without increasing the overall size of the police force.

…

Why do police departments allocate their officers so inefficiently? Devi and Fryer suggest several possible explanations, such as political pressure, union rules, historical inertia, or lack of data. They also point out some potential barriers to implementing a more efficient allocation, such as officer preferences, community resistance, or legal constraints.

11) Really great from NYT, “Does Therapy Really Work? Let’s Unpack That.”

The answer is complicated. The research shows that therapy does work for many people — but not for everyone. And it’s hard to say exactly what kind of therapy works best for whom, or under what circumstances. The effectiveness of therapy depends on many factors, such as the type and severity of the problem, the quality of the therapist-client relationship, the client’s motivation and expectations, and the therapist’s training and experience.

…

One way to measure the effectiveness of therapy is to use meta-analyses, which combine the results of many studies on the same topic. Meta-analyses can provide an overall estimate of how much therapy helps people improve their mental health, compared with not receiving any treatment or receiving a placebo.

…

According to a 2018 meta-analysis by Pim Cuijpers and colleagues, which included 421 studies with more than 36,000 participants, the average effect size of therapy was 0.69. This means that after receiving therapy, the average client was better off than 76 percent of people who did not receive therapy.

…

Another way to measure the effectiveness of therapy is to use benchmarks, which compare the outcomes of therapy with those of other treatments or natural recovery. Benchmarks can help answer the question: How much better off are people who receive therapy than people who receive other forms of help or no help at all?

…

According to a 2013 meta-analysis by Bruce Wampold and Zac Imel, which included 79 studies with more than 7,000 participants, the average effect size of therapy compared with benchmarks was 0.51. This means that after receiving therapy, the average client was better off than 69 percent of people who received other forms of help or no help at all.

12) Interesting stuff in the Lancet on how to think about obesity:

Oooof– summarized a different Lancet article!!  I’ll have to do it myself. 

In practical terms, this definition requires the health professional to answer the following question: Does this patient present with a health problem that is likely to improve with weight loss? If the answer is “yes”, then the patient has obesity. If not, then the patient may just have adiposity, which may well at some stage progress to overt obesity (hence the suggestion to refer to these individuals as having pre-obesity).

 

Such an approach to diagnosing obesity would of course require a clinical assessment of each patient by a qualified health practitioner. Only a comprehensive interview together with a physical exam as well as relevant laboratory and imaging tests would establish (or rule out) the diagnosis “obesity” in a given individual. While this clearly makes the diagnosis of obesity more cumbersome, it ensures that otherwise healthy individuals are no longer labeled as having obesity simply based on their size. Perhaps, more importantly, individuals presenting with health issues that are clearly linked to or likely to improve with weight loss, can be diagnosed with having obesity (and thus qualifying for obesity treatments), even when they fall below the conventional BMI cutoffs. While this introduces an element of clinical judgment into the diagnosis, this is not uncommon in medical practice, where clinical judgment is often called upon in determining the presence and severity of a medical issue and the best course of action.

 

Ultimately, the goal of making a proper diagnosis is to determine the right course of action for a given individual. In the case of someone presenting with a health problem closely linked to excess weight, for which we have strong evidence that weight-loss would improve it (e.g. hypertension, type 2 diabetes, obstructive sleep apnoea, etc.), we would see a “primary” indication for obesity treatment, i.e. successful reduction in body weight can essentially solve the problem (Fig. 1). However, we may also be confronted with a patient who presents with a health problem, not causally linked to obesity, but which is aggravated by or more difficult to manage due to the presence of excess weight (e.g. someone with excess weight who sustains an injury or contracts COVID). Such an individual could be considered to have a “secondary” indication for obesity treatment. While weight-loss will not solve the underlying problem, it may make management and recovery easier. Finally, we may consider individuals with excess weight, who present with a health problem that is neither related to nor likely to improve with weight loss. This person may be considered to have a “tertiary” indication for obesity treatment, which although perhaps leading to an overall improvement in health, would have no impact on the presenting complaint.

13) Ross Douthat’s case against legalizing marijuana didn’t strike me as particularly strong:

Of all the ways to win a culture war, the smoothest is to just make the other side seem hopelessly uncool. So it’s been with the march of marijuana legalization: There have been moral arguments about the excesses of the drug war and medical arguments about the potential benefits of pot, but the vibe of the whole debate has pitted the chill against the uptight, the cool against the square, the relaxed future against the Principal Skinners of the past.

…

All of this means that it will take a long time for conventional wisdom to acknowledge the truth that seems readily apparent to squares like me: Marijuana legalization as we’ve done it so far has been a policy failure, a potential social disaster, a clear and evident mistake.

…

The best version of the square’s case is an essay by Charles Fain Lehman of the Manhattan Institute explaining his evolution from youthful libertarian to grown-up prohibitionist. It will not convince readers who come in with stringently libertarian presuppositions — who believe on high principle that consenting adults should be able to purchase, sell and enjoy almost any substance short of fentanyl and that no second-order social consequence can justify infringing on this right. But Lehman explains in detail why the second-order effects of marijuana legalization have mostly vindicated the pessimists and skeptics.

…

First, on the criminal justice front, the expectation that legalizing pot would help reduce America’s prison population by clearing out nonviolent offenders was always overdrawn, since marijuana convictions made up a small share of the incarceration rate even at its height. But Lehman argues that there is also no good evidence so far that legalization reduces racially discriminatory patterns of policing and arrests.

I like this Dilan Esper response:

Since there's been a lot of pushback recently from smart people against marijuana legalization, I think it's worth making a couple of points.

First, and most importantly, Prohibition was a failure, just as it was with alcohol. 1/https://t.co/THQEdnW2E1

— Dilan Esper (@dilanesper) May 19, 2023

14) The WHO’s case against artificial sweeteners is even less compelling. Also, the WHO, of course, is the organization that was insisting on droplet transmission of Covid a whole damn year after everyone else knew it was airborne.

If you’re trying to lose weight or prevent weight gain, products sweetened with artificial sweeteners rather than with higher calorie table sugar may be an attractive option. Artificial sweeteners are many times sweeter than table sugar, so smaller amounts are needed to create the same level of sweetness.

…

But do artificial sweeteners actually help reduce calories and deliver on their promise to help you lose weight? A new report from the World Health Organization suggests that they don’t.

…

The report, published on Monday in The BMJ, is based on a systematic review of 56 studies that examined the effects of non-sugar sweeteners on health outcomes in both adults and children. The researchers found that there was no compelling evidence to indicate that artificial sweeteners help people lose weight over time. Nor did they find any clear evidence that they prevent obesity or other conditions such as diabetes, cancer and dental decay.

…

The researchers did find some evidence that artificial sweeteners may have a modest benefit for reducing body mass index and fasting blood glucose levels. But they said these findings were based on low-quality studies with a high risk of bias, and that more research is needed to confirm them.

…

The report also noted that there are many uncertainties about the potential harms of artificial sweeteners. Some studies have suggested that they may alter the gut microbiota and affect appetite and glucose regulation. Other studies have raised concerns about possible links between artificial sweeteners and cancer, cardiovascular disease and kidney damage.

15) Scott Alexander on the weirdness of the academic job market:

The academic job market is weird. It’s weird in a way that’s hard to explain to people who haven’t experienced it. It’s weird in a way that makes it hard for people who are in it to make rational decisions.

…

The weirdness starts with the fact that academic jobs are scarce and highly competitive. There are far more PhDs than there are tenure-track positions, and getting one of those positions requires not only years of training and research, but also luck, timing, networking, and strategic choices.

…

The weirdness continues with the fact that academic jobs are highly specialized and geographically dispersed. Unlike most other professions, where you can apply for jobs in your field in different cities or regions, academic jobs are tied to specific departments and disciplines. You can’t just decide to move to a new place and look for a job there; you have to wait for a job opening that matches your expertise and interests, and hope that it’s in a location that you like or can tolerate.

…

The weirdness culminates with the fact that academic jobs are highly uncertain and contingent. Even if you get a tenure-track position, you still have to go through a probationary period of several years, during which you have to prove yourself by publishing, teaching, and securing grants. If you fail to meet the expectations of your department or university, you can be denied tenure and lose your job. And even if you get tenure, you still have to deal with the pressures and challenges of academia, such as increasing workloads, shrinking budgets, changing student demographics, and shifting intellectual trends.¹