Persuasion, the process of altering someone’s belief, position or opinion on a specific matter, is pervasive in human affairs and a widely studied topic in the social sciences. From public health campaigns to marketing and sales to political propaganda, various actors develop elaborate persuasive communication strategies on a large scale, investing substantial resources to make their messaging resonate with broad audiences. In recent decades, the diffusion of social media and other online platforms has expanded the potential of mass persuasion by enabling personalization or ‘microtargeting’—the tailoring of messages to an individual or a group to enhance their persuasiveness. The efficacy of microtargeting has been questioned because it relies on the assumption of effect heterogeneity, that is, that specific groups of people respond differently to the same inputs, a concept that has been disputed in previous literature. Nevertheless, microtargeting has proven effective in a variety of settings, and most scholars agree on its persuasive power.

Microtargeting practices are fundamentally constrained by the burden of profiling individuals and crafting personalized messages that appeal to specific targets, as well as by a restrictive interaction context without dialogue. These limitations may soon fall off due to the recent rise of large language models (LLMs)—machine learning models trained to mimic human language and reasoning by ingesting vast amounts of textual data.

In the context of persuasion, experts have widely expressed concerns about the risk of LLMs being used to manipulate online conversations and pollute the information ecosystem by spreading misinformation, exacerbating political polarization, reinforcing echo chambers and persuading individuals to adopt new beliefs. This is especially relevant since LLMs and other AI systems are capable of inferring personal attributes from publicly available digital traces such as Facebook likes, status updates and messages, Reddit and Twitter posts, pictures liked on Flickr, and other digital footprints. In addition, users find it increasingly challenging to distinguish AI-generated from human-generated content, with LLMs efficiently mimicking human writing and thus gaining credibility. […]

Our results show that, on average, GPT-4 opponents outperformed human opponents across every topic and demographic, exhibiting a high level of persuasiveness. In particular, when compared to the baseline condition of debating with a human, debating with GPT-4 with personalization resulted in a +81.2% increase in the odds of reporting higher agreements with opponents. More intuitively, this means that 64.4% of the time, personalized GPT-4 opponents were more persuasive than human opponents. […]

In other words, not only was GPT-4 able to exploit personal information to tailor its arguments effectively, but it also succeeded in doing so far more effectively than humans.

Our study suggests that concerns around personalization and AI persuasion are warranted […] We emphasize that the effect of personalization is particularly remarkable given how little personal information was collected (gender, age, ethnicity, education level, employment status and political affiliation) and despite the extreme simplicity of the prompt instructing the LLM to incorporate such information. […]

A promising approach to counter mass disinformation campaigns could be enabled by LLMs themselves, generating similarly personalized counternarratives to educate bystanders potentially vulnerable to deceptive posts.

{ Nature | Continue reading }

AI has poisoned its own well

Replied to The Curse of Recursion: Training on Generated Data Makes Models Forget (arXiv.org)

What will happen to GPT-{n} once LLMs contribute much of the language found online? We find that use of model-generated content in training causes irreversible defects in the resulting models, where tails of the original content distribution disappear. […] the value of data collected about genuine human interactions with systems will be increasingly valuable in the presence of content generated by LLMs in data crawled from the Internet.

I suspect tech companies (particularly Microsoft / OpenAI and Google) have miscalculated, and in their fear of being left behind, have released their generative AI models too early and too wide. By doing so, they’ve essentially established a threshold for the maximum improvement of their products due to the threat of model collapse.[…]

They need an astronomical amount of training data to make any model better than what already exists. By releasing their models for public use now, when they’re not very good yet, too many people have pumped the internet full of mediocre generated content with no indication of provenance. […]

Obtaining quality training data is going to be very expensive in five years if AI doesn’t win all its lawsuits over training data being fair use.

{ Tracy Durnell | Continue reading }

The possibility of points-of-no-return in the climate system has been discussed for two decades. A point-of-no-return can be seen as a threshold which, once surpassed, fundamentally changes the dynamics of the climate system. For example, by triggering irreversible processes like melting of the permafrost, drying of the rainforests, or acidification of surface waters. Recently, Lenton et al. summarized the global situation and warned that thresholds may be closer in time than commonly believed.

The purpose of this article is to report that we have identified a point-of-no-return in our climate model—and that it is already behind us. ESCIMO is a climate model which we run from 1850 to 2500. In ESCIMO the global temperature keeps rising to 2500 and beyond, irrespective of how fast humanity cuts the emissions of man-made greenhouse gas (GHG) emissions. […]

To stop the self-sustained warming in ESCIMO, enormous amounts of CO₂ have to be extracted from the atmosphere.

{ Nature | Continue reading }

oil on canvas { John William Waterhouse, Pandora, 1898 }

Most estimates suggest 80% of COVID-19 cases are mild and feel roughly like a flu. Estimates I have seen suggest that roughly 10-15% of cases will be more significant and may necessitate hospital visits (see also) with 1-3% potentially needing an ICU. The concern of many governments is the peak number of cases that occur in a given moment. […]

The reported death rate has hovered around 2% but may in reality be 0.2% to 1% depending on country and healthcare system. Many estimates tend to indicate an overall expected mortality rate of ~0.5% globally. The current existing fatality rate is biased upwards by Wuhan cases dominating the mix (which are closer to a 3-4% death rate and make up most cases). It is possible the virus is being undertested for in China / rest of world driving the real death rate down (as many more people are infected than is reported). […]

R0 value: The spread rate of the virus seems to be well over 2 and likely ~3. This means for every person infected at least 2 to 3 more get the disease.

{ EladGil | Continue reading }

Experts think there may be many people with no symptoms at all, or such mild ones that they never bother to seek medical attention. Because those cases have not been counted, it’s not possible now to know the real proportion of mild versus severe cases. […]

After viral infections, people generally develop antibodies in their blood that will fight off the virus and protect them from contracting it again. It’s reasonable to assume that people who have had the new coronavirus will become immune to it.

But it is not known how long that immunity will last. With other coronaviruses, which cause the common cold, immunity can wane.

{ NY Times | Continue reading }

The best estimates so far suggest that Covid-19 kills about 1% of people it infects. That number may go up somewhat or fall significantly; either way it could add up to a dreadful toll.

If 60% of the world’s population is ultimately infected, as suggested by Gabriel Leung, chair of public health medicine at Hong Kong University, a 1% fatality rate would kill almost 50 million people — similar to the 1918 Spanish flu. If that falls to 0.1%, it could still be roughly 10 times more fatal than the 2009 H1N1 influenza outbreak, which killed several hundred thousand in its first year. […]

The most severe period of initial infection could soon be fading. Respiratory diseases flourish in the cold season and taper off as the weather warms up. That should cause infection rates to slow in the northern hemisphere, while continuing at a lower level in tropical regions and spiking in temperate parts of the southern hemisphere where winter will be setting in. When a new year rolls around, the bulk of the disease will shift back to the northern hemisphere, to begin the cycle again.

Subsequent Covid-19 seasons probably won’t be as serious. Those who survive viruses should be immune from reinfection (though there have been reports of people being diagnosed with Covid-19 for a second time), and as the share of survivors in the population rises, it gets harder for a disease to spread. […]

In a best-case scenario, it’s even possible that vaccines may be available in not much more than a year.

{ Bloomberg | Continue reading }

The only path to flattening the curve for COVID-19 is community-wide isolation: the more people stay home, the fewer people will catch the disease. The fewer people who catch the disease, the better hospitals can help those who do. […]

Get a flu shot, if you haven’t already, and stock up supplies at home so that you can stay home for two or three weeks, going out as little as possible. […] Here’s a handy, one-page guide on what you need.

{ Scientific American | Continue reading }

related { CoronaCoin: crypto developers seize on coronavirus for new, morbid token }

Old emails, photos and files from years past sit undisturbed, awaiting a search […] The problem is that all those messages require energy to preserve them. […]

Right now, data centers consume about 2 percent of the world’s electricity, but that is expected to reach 8 percent by 2030. Moreover, only about 6 percent of all data ever created is in active use today, according to research from Hewlett Packard Enterprise. That means 94 percent is sitting in a vast “landfill” with a massive carbon footprint.

“It’s costing us the equivalent of maintaining the airline industry for data we don’t even use”

{ Japan Times | Continue reading }

Ethos Capital, a new commercial investment firm founded in the past few months in Boston, has 2 staff and only one major investment: a deal to acquire the 501c3 non-profit [Public Interest Registry] that currently runs the .org domain (valued at a few $B), for an undisclosed sum.

This was initiated immediately after ICANN decided in May, over almost universal opposition, to remove the price cap on .org registrations with no meaningful price protections for existing or future registrants.

{ The Longest Now | Continue reading }

Internet Society (ISOC) has sold the .org registry Public Interest Registry (PIR) to private equity company Ethos Capital. […] PIR generated $101 million in revenue in 2018 and contributed nearly $50 million to Internet Society. […]

Ethos Capital is a new private equity firm lead by Erik Brooks. Brooks was at Abry Partners until earlier this year. Abry Partners acquired Donuts and installed former ICANN President of Global Domains Akram Atallah in the top spot there. […] The other person at Ethos is former ICANN Senior Vice President Abusitta-Ouri.

{ Domain Name Wire | Continue reading }

{ Horns are growing on young people’s skulls, caused by the forward tilt of the head. Phone use is to blame, research suggests. | Washington Post | No, Teenagers Are Not Growing ‘Skull Horns’ Because of Smartphones | Time }

We fail to see that a computer that is a hundred times more accurate than a human, and a million times faster, will make 10,000 times as many mistakes.

{ The Guardian | Continue reading }

related { Robots Will Replace Doctors, Lawyers, and Other Professionals }

art { electropollock | video }

Since 1955, The Journal of Irreproducible Results has offered “spoofs, parodies, whimsies, burlesques, lampoons and satires” about life in the laboratory. Among its greatest hits: “Acoustic Oscillations in Jell-O, With and Without Fruit, Subjected to Varying Levels of Stress” and “Utilizing Infinite Loops to Compute an Approximate Value of Infinity.” The good-natured jibes are a backhanded celebration of science. What really goes on in the lab is, by implication, of a loftier, more serious nature.

It has been jarring to learn in recent years that a reproducible result may actually be the rarest of birds. Replication, the ability of another lab to reproduce a finding, is the gold standard of science, reassurance that you have discovered something true. But that is getting harder all the time. With the most accessible truths already discovered, what remains are often subtle effects, some so delicate that they can be conjured up only under ideal circumstances, using highly specialized techniques.

Fears that this is resulting in some questionable findings began to emerge in 2005, when Dr. John P. A. Ioannidis, a kind of meta-scientist who researches research, wrote a paper pointedly titled “Why Most Published Research Findings Are False.”

[…]

The fear that much published research is tainted has led to proposals to make replication easier by providing more detailed documentation, including videos of difficult procedures. […] Scientists talk about “tacit knowledge,” the years of mastery it can take to perform a technique. The image they convey is of an experiment as unique as a Rembrandt.

{ NY Times | Continue reading }

This report spells out what the world would be like if it warmed by 4 degrees Celsius, which is what scientists are nearly unanimously predicting by the end of the century, without serious policy changes.

The 4°C scenarios are devastating: the inundation of coastal cities; increasing risks for food production potentially leading to higher malnutrition rates; many dry regions becoming dryer, wet regions wetter; unprecedented heat waves in many regions, especially in the tropics; substantially exacerbated water scarcity in many regions; increased frequency of high-intensity tropical cyclones; and irreversible loss of biodiversity, including coral reef systems. […]

The science is unequivocal that humans are the cause of global warming, and major changes are already being observed: global mean warming is 0.8°C above pre industrial levels; oceans have warmed by 0.09°C since the 1950s and are acidifying; sea levels rose by about 20 cm since pre-industrial times and are now rising at 3.2 cm per decade; an exceptional number of extreme heat waves occurred in the last decade; major food crop growing areas are increasingly affected by drought.

{ World Bank | PDF }

China is a kleptocracy of a scale never seen before in human history. This post aims to explain how  this wave of theft is financed, what makes it sustainable and what will make it fail.

{ Bronte Capital | Continue reading }

artwork { Li Shida }

Using scientific theories, toy ecosystem modeling and paleontological evidence as a crystal ball, 18 scientists, including one from Simon Fraser University, predict we’re on a much worse collision course with Mother Nature than currently thought. […]

Earth’s accelerating loss of biodiversity, its climates’ increasingly extreme fluctuations, its ecosystems’ growing connectedness and its radically changing total energy budget are precursors to reaching a planetary state threshold or tipping point.

Once that happens, which the authors predict could be reached this century, the planet’s ecosystems, as we know them, could irreversibly collapse in the proverbial blink of an eye.

{ EurekAlert | Continue reading }

A massive, highly sophisticated piece of malware has been newly found infecting systems in Iran and elsewhere and is believed to be part of a well-coordinated, ongoing, state-run cyberespionage operation. […] Dubbed “Flame” by Russia-based anti-virus firm Kaspersky Lab […]

The malware, which is 20 megabytes when all of its modules are installed, contains multiple libraries, SQLite3 databases, various levels of encryption — some strong, some weak — and 20 plug-ins that can be swapped in and out to provide various functionality for the attackers. It even contains some code that is written in the LUA programming language — an uncommon choice for malware.

Kaspersky Lab is calling it “one of the most complex threats ever discovered.” […]

Gostev says that because of its size and complexity, complete analysis of the code may take years. “It took us half-a-year to analyze Stuxnet,” he said. “This is 20-times more complicated. It will take us 10 years to fully understand everything.” […]

Among Flame’s many modules is one that turns on the internal microphone of an infected machine to secretly record conversations that occur either over Skype or in the computer’s near vicinity; a module that turns Bluetooth-enabled computers into a Bluetooth beacon, which scans for other Bluetooth-enabled devices in the vicinity to siphon names and phone numbers from their contacts folder; and a module that grabs and stores frequent screenshots of activity on the machine, such as instant-messaging and email communications, and sends them via a covert SSL channel to the attackers’ command-and-control servers.

The malware also has a sniffer component that can scan all of the traffic on an infected machine’s local network and collect usernames and password hashes that are transmitted across the network. The attackers appear to use this component to hijack administrative accounts and gain high-level privileges to other machines and parts of the network. […]

Because Flame is so big, it gets loaded to a system in pieces. The machine first gets hit with a 6-megabyte component, which contains about half-a-dozen other compressed modules inside. The main component extracts, decompresses and decrypts these modules and writes them to various locations on disk. The number of modules in an infection depends on what the attackers want to do on a particular machine.

Once the modules are unpacked and loaded, the malware connects to one of about 80 command-and-control domains to deliver information about the infected machine to the attackers and await further instruction from them. The malware contains a hardcoded list of about five domains, but also has an updatable list, to which the attackers can add new domains if these others have been taken down or abandoned.

While the malware awaits further instruction, the various modules in it might take screenshots and sniff the network. The screenshot module grabs desktop images every 15 seconds when a high-value communication application is being used, such as instant messaging or Outlook, and once every 60 seconds when other applications are being used.

{ Wired | Continue reading }

The outbreak of a new livestock disease in western Europe last year, particularly harmful to offspring, could move further into areas surrounding the worst affected countries in the next cycle of new births, scientists say.

Schmallenberg virus - named after the German town where it was first detected in November - infected sheep and cows on at least 2,600 farms in eight EU countries last year, most likely between August and October.

Thought to have been spread for hundreds of miles across Europe by biting midges and warm late summer winds, the virus has since been confirmed in Belgium, the Netherlands, Luxembourg, France, Italy, Spain and Britain.

{ Reuters | Continue reading }

Computers dominate how we live, work and think. For some, the technology is a boon and promises even better things to come. But others warn that there could be bizarre consequences and that humans may be on the losing end of progress.  (…)

“Economic progress ultimately signifies the ability to produce things at a lower financial cost and with less labor than in the past,” says Polish sociologist Zygmunt Bauman. As a result, he says, increasing effectiveness goes hand in hand with rising unemployment, and the unemployed merely become “human waste.”

Likewise, (…) Erik Brynjolfsson and Andrew McAfee, both scholars at the MIT, argue that, for the first time in its history, technological progress is creating more jobs for computers than for people.

{ Spiegel | Continue reading }

unrelated { Competition among memes in a world with limited attention }

As bacteria evolve to evade antibiotics, common infections could become deadly, according to Dr. Margaret Chan, director general of the World Health Organization.

Speaking at a conference in Copenhagen, Chan said antibiotic resistance could bring about “the end of modern medicine as we know it.”

“We are losing our first-line antimicrobials,” she said Wednesday in her keynote address at the conference on combating antimicrobial resistance. “Replacement treatments are more costly, more toxic, need much longer durations of treatment, and may require treatment in intensive care units.”

Chan said hospitals have become “hotbeds for highly-resistant pathogens” like methicillin-resistant Staphylococcus aureus, “increasing the risk that hospitalization kills instead of cures.”

Indeed, diseases that were once curable, such as tuberculosis, are becoming harder and more expensive to treat.

{ ABC | Continue reading }

artwork { Mœbius }

We’ve learned that Mike Daisey’s story about Apple in China - which we broadcast in January - contained significant fabrications. We’re retracting the story because we can’t vouch for its truth. This is not a story we commissioned. It was an excerpt of Mike Daisey’s acclaimed one-man show “The Agony and the Ecstasy of Steve Jobs,” in which he talks about visiting a factory in China that makes iPhones and other Apple products.

{ This American Life | Continue reading | More: Mike Daisey’s Lies About China }

How would things change if Google and Bing went down for 24 hours, and there wasn’t a way around the block?

If your first thought is to do your online searches through Yahoo!, you will run into another roadblock. Since 2010, Yahoo! searches are powered by Bing. Can you name any other search engine sites off the top of your head? (…)

Losing search sites is only the tip of the iceberg. Google and Bing also provide extensive services in other areas, one of the most obvious being email—Gmail alone has 350 million users. Blacking out Gmail would certainly affect all these people, but it would also affect everyone trying to reach them.

{ Naked Capitalism | Continue reading }

Last week, I got a notice from Twitter saying the Manhattan District Attorney’s office had subpoenaed my account activity for a three-month period between September and December of last year. On October 1, I was arrested along with 700 or so other people marching across the Brooklyn Bridge as part of an early Occupy Wall Street demonstration. (…)

Why was it Twitter who got subpoenaed even though they’re my words the DA wants to see?

The short answer is: they’re not my words. Not in the legal sense at least. Part of the Twitter user agreement is that the Tweets belong to the company, not to the user. As far as the law is concerned, my online self is an informational aspect of a legal entity named Twitter, not me. That means if someone wants to use my statements against me in court, it’s not me they have to call, it’s that little blue birdie. In this context the term “microblogging” gets some new meaning: Twitter’s users really are unpaid content producers for a giant microblog hosting site.

{ Malcolm Harris/Shareable | Continue reading }

related { Will the Web Break? }

photo { Guy Bourdin }

{ Apple sells more phones in a day than people make babies | An hour of video posted every second on YouTube }

image { Robert Mangold, 1/3 Gray-Green Curved Area, 1966 | Guggenheim, until Feb. 8 }

Climate scientists have long warned that if we continue to burn fossil fuels like coal, oil and natural gas as our dominant source of energy, the planet will warm, extreme events will increase, and we will become more vulnerable to disasters. Overall, the planet has warmed about 1.2°F over the past century. Since I was born in 1970, the United States has heated up at a pace of 0.5°F per decade. As Lemonick points out, ”Scientists know that the increasing load of greenhouse gases we’re pumping into the atmosphere doesn’t “cause” extreme weather. But it does raise the odds, just as a diet of triple bacon cheeseburgers raises the odds of heart disease.”

All weather is now born into an environment that is warmer and moister because of man-made, heat-trapping greenhouse gas pollution. Thanks in part to warmer oceans, there is 4 percent more water vapor in the atmosphere and that amount will continue to increase as the planet warms, providing more fuel for storms. Droughts, wildfires, heat waves and heavy downpours are going to become more frequent, more intense and longer lasting. In fact, we can already see this playing out in historical data.

{ Salon | Continue reading | More: Why do people still deny climate change? }

As the Intergovernmental Panel on Climate Change concluded in its most recent assessment report: “warming of the climate system is unequivocal, as is now evident from observations of increases in global average air and ocean temperatures, widespread melting of snow and ice, and rising global average sea level.” And as data continue to pile up, the evidence gets ever stronger that human-induced emissions of greenhouse gases are the main cause of the observed warming over the past century.

But hurricanes are difficult. Climate models predict that they will become more intense. At the same time, considerable uncertainty remains. We have only about 40 years of reliable observational records, which precludes a clear determination of their variability. Given that different aspects of climate change could act to increase or decrease hurricane activity, whether or not Katrina can be ascribed to global warming is a challenge beset by difficulty.

{ Slate | Continue reading | More: When should we blame climate change for natural disasters? }

Humanity has done little to address climate change. Global emissions of carbon dioxide reached (another) all-time peak in 2010. The most recent international talks to craft a global treaty to address the problem pushed off major action until 2020. Fortunately, there’s an alternative—curbing the other greenhouse gases.

Specifically, in the case of rapid action to slow catastrophic climate change, the best alternatives appear to be: methane and black carbon (otherwise known as soot). A new economic and scientific analysis published in Science on January 13 of the benefits of cutting these two greenhouse gases finds the benefits to be manifold—from human health to increased agricultural yields.

Even better, by analyzing some 400 potential soot- and methane-emission control measures, the international team of researchers found that just 14 deliver “nearly 90 percent” of the potential benefits. Bonus: the 14 steps also restrain global warming by roughly 0.5 degree Celsius by 2050, according to computer modeling.

{ Scientific American | Continue reading }