Researchers widely observe yet seldom publish about same-sex sexual behavior in primates and other mammals – often because it is perceived to be rare. Credit: Selvan Tamilmani, Unsplash, CC0 (creativecommons.org/publicdomain/zero/1.0/)
A team of anthropologists and biologists from Canada, Poland, and the U.S., working with researchers at the American Museum of Natural History, in New York, has found via meta-analysis of data from prior research efforts that homosexual behavior is far more common in other animals than previously thought. The paper is published in PLOS ONE.
For many years, the biology community has accepted the notion that homosexuality is less common in animals than in humans, despite a lack of research on the topic. In this new effort, the researchers sought to find out if such assumptions are true.
The work involved study of 65 studies into the behavior of multiple species of animals, mostly mammals, such as elephants, squirrels, monkeys, rats and racoons.
The researchers found that 76% of the studies mentioned observations of homosexual behavior, though they also noted that only 46% had collected data surrounding such behavior—and only 18.5% of those who had mentioned such behavior in their papers had focused their efforts on it to the extent of publishing work with homosexuality as it core topic.
They noted that homosexual behavior observed in other species included mounting, intromission and oral contact—and that researchers who identified as LGBTQ+ were no more or less likely to study the topic than other researchers.
The researchers point to a hesitancy in the biological community to study homosexuality in other species, and thus, little research has been conducted. They further suggest that some of the reluctance has been due to the belief that such behavior is too rare to warrant further study.
The research team suggests that homosexuality is far more common in the animal kingdom than has been reported—they further suggest more work is required regarding homosexual behaviors in other animals to dispel the myth of rarity.
More information:
Karyn A. Anderson et al, Same-sex sexual behaviour among mammals is widely observed, yet seldomly reported: Evidence from an online expert survey, PLOS ONE (2024). DOI: 10.1371/journal.pone.0304885
Citation:
Analysis of data suggests homosexual behavior in other animals is far more common than previously thought (2024, June 24)
retrieved 25 June 2024
from https://phys.org/news/2024-06-analysis-homosexual-behavior-animals-common.html
This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no
part may be reproduced without the written permission. The content is provided for information purposes only.
Income difference in percent if parents have a degree in the same field of study (selection). Credit: The British Journal of Sociology (2024). DOI: 10.1111/1468-4446.13090
Lawyers, doctors and engineers who have studied in the same field as their parents have higher incomes than their colleagues, according to a new Danish study published in The British Journal of Sociology. While this could be a sign of nepotism or other forms of discrimination, the explanation is likely to be something else entirely.
“Nepo baby” is the less flattering term for children who get jobs in the same field as their parents, helped by their parents’ connections and reputation. The children are given a head start in a working life where the route to success is not just about ability and talent.
In the cultural or media world in particular, “nepo babies” provoke debate. But in a broader and more general sense, is it also an advantage to have the same education as one’s parents? And if so, what are the mechanisms at work?
Assistant Professor Jesper Fels Birkelund, from the Department of Sociology, explores these questions in a new study. He maps the income of 40-year-olds who have followed their father’s or mother’s educational path and compares it with the many who have not.
The study examines the earnings advantage of having a degree in the same field of study as one’s parents. The main findings are based on extensive statistical analyses of register data on 139,000 Danes born between 1960 and 1979. Everyone in the sample attended academic upper secondary education and—like their parents—obtained at least a bachelor’s degree, including a professional bachelor’s degree.
Using a special methodological approach, the study also compares the income of siblings who did or did not choose the same education as their parents. This is done to rule out family-specific explanations for income differences—for example, that families in which a child follows in its parents’ footsteps are more resourceful or characterized by stronger emotional ties between parents and children.
The extensive statistical analysis concludes that there is generally a small financial advantage to having the same higher education as one’s parents. However, the study also documents that the picture varies greatly according to educational background.
“The association is most clear for medicine, law and, to some extent, engineering. Doctors or lawyers earn on average 5%–10% more if at least one of their parents has a degree in the same field of study. However, these significant differences are only found in a few professions. For many public sector employees, such as nurses and teachers, salaries are so regulated that there is no effect.”
The differences between fields also mean that the overall earnings effect of following in one’s parents’ footsteps is smaller. Across all areas of education, the effect is around 2%. At the same time, however, the new study adds nuance to other studies that have shown that an individual’s social background becomes less important for their labor market success once they have a bachelor’s degree.
No signs of nepotism
But what explains why children of doctors, lawyers and engineers seem to benefit financially from having the same education as their parents?
According to Fels Birkelund, the evidence points to a particular factor, especially for the first two groups. “Relatively many doctors and lawyers have income from self-employment, for example from a private medical or legal practice. And the differences in income are largely due to the higher income from this source.
“So children who follow in their parents’ footsteps are more likely than others to be successfully self-employed.”
And they seem to get by without direct help from their parents. There is nothing in the study to suggest that the children have benefited from any special privileges or actual nepotism: They rarely take over their parents’ businesses. And they are not more likely to be employed in companies where people from their parents’ network hold senior management positions.
“The study thereby excludes some mechanisms that are typically brought up in the ‘nepotism’ debate. For example, that children use their parents’ social networks to get a job,” says Fels Birkelund.
Benefits from the upbringing
Because of its broad design, the study cannot map nepotism in small niches such as the cultural and media world. Nor can it identify the general mechanisms that make some people better at generating income through their own business or otherwise.
However, Fels Birkelund points out that children generally receive some generic skills from their upbringing and parents. A human and cultural “capital” that helps them later in life.
“They may have been instilled with a particular entrepreneurial spirit or learned to thrive in certain professional environments. They may also have developed certain skills while growing up. For example, you could imagine that the children of engineers might have played more with technical toys than others. The study can’t pinpoint the exact mechanisms, but it suggests that there’s more at play than just nepotism.”
More information:
Jesper Fels Birkelund, Economic returns to reproducing parents’ field of study, The British Journal of Sociology (2024). DOI: 10.1111/1468-4446.13090
Citation:
Following in parents’ educational footsteps offers financial rewards, study shows (2024, June 24)
retrieved 25 June 2024
from https://phys.org/news/2024-06-parents-footsteps-financial-rewards.html
This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no
part may be reproduced without the written permission. The content is provided for information purposes only.
The “SnailLoad” loophole is based on combining the latency of internet connections with the fingerprinting of online content. Credit: IAIK – TU Graz
Internet users leave many traces on websites and online services. Measures such as firewalls, VPN connections and browser privacy modes are in place to ensure a certain level of data protection. However, a newly discovered security loophole allows bypassing all of these protective measures.
Computer scientists from the Institute of Applied Information Processing and Communication Technology (IAIK) at Graz University of Technology (TU Graz) were able to track users’ online activities in detail simply by monitoring fluctuations in the speed of their internet connection. No malicious code is required to exploit this vulnerability, known as “SnailLoad,” and the data traffic does not need to be intercepted. All types of end devices and internet connections are affected.
The researchers have published their work in a paper titled “SnailLoad: Exploiting Remote Network Latency Measurements without JavaScript.”
Attackers track latency fluctuations in the internet connection via file transfer
Attackers only need to have had direct contact with the victim on a single occasion beforehand. On that occasion, the victim downloads a basically harmless, small file from the attacker’s server without realizing it—for example, while visiting a website or watching an advertising video.
As this file does not contain any malicious code, it cannot be recognized by security software. The transfer of this file is extremely slow, providing the attacker with continuous information about the latency variation of the victim’s internet connection. In further steps, this information is used to reconstruct the victim’s online activity.
‘SnailLoad’ combines latency data with fingerprinting of online content
“When the victim accesses a website, watches an online video or speaks to someone via video, the latency of the internet connection fluctuates in a specific pattern that depends on the particular content being used,” says Stefan Gast from the IAIK. This is because all online content has a unique fingerprint: For efficient transmission, online content is divided into small data packages that are sent one after the other from the host server to the user. The pattern of the number and size of these data packages is unique for each piece of online content—like a human fingerprint.
The researchers collected the fingerprints of a limited number of YouTube videos and popular websites in advance for testing purposes. When the test subjects used these videos and websites, the researchers were able to recognize this through the corresponding latency fluctuations.
“However, the attack would also work the other way round,” says Daniel Gruss from the IAIK. “Attackers first measure the pattern of latency fluctuations when a victim is online and then search for online content with the matching fingerprint.”
Slow internet connections make it easier for attackers
When spying on test subjects who were watching videos, the researchers achieved a success rate of up to 98%.
“The higher the data volume of the videos and the slower the victims’ internet connection, the better the success rate,” says Gruss. Consequently, the success rate for spying on basic websites dropped to around 63%.
“However, if attackers feed their machine learning models with more data than we did in our test, these values will certainly increase,” says Gruss.
Loophole virtually impossible to close
“Closing this security gap is difficult. The only option would be for providers to artificially slow down their customers’ internet connections in a randomized pattern,” says Gruss. However, this would lead to noticeable delays for time-critical applications such as video conferences, live streams or online computer games.
Citation:
New security loophole allows spying on internet users visiting websites and watching videos (2024, June 24)
retrieved 25 June 2024
from https://techxplore.com/news/2024-06-loophole-spying-internet-users-websites.html
This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no
part may be reproduced without the written permission. The content is provided for information purposes only.
The proposed onboarding approach with the IntegrAI algorithm. Credit: arXiv (2023). DOI: 10.48550/arxiv.2311.01007
Artificial intelligence models that pick out patterns in images can often do so better than human eyes—but not always. If a radiologist is using an AI model to help her determine whether a patient’s X-rays show signs of pneumonia, when should she trust the model’s advice and when should she ignore it?
A customized onboarding process could help this radiologist answer that question, according to researchers at MIT and the MIT-IBM Watson AI Lab. They designed a system that teaches a user when to collaborate with an AI assistant.
In this case, the training method might find situations where the radiologist trusts the model’s advice—except she shouldn’t because the model is wrong. The system automatically learns rules for how she should collaborate with the AI, and describes them with natural language.
During onboarding, the radiologist practices collaborating with the AI using training exercises based on these rules, receiving feedback about her performance and the AI’s performance.
The researchers found that this onboarding procedure led to about a 5 percent improvement in accuracy when humans and AI collaborated on an image prediction task. Their results also show that just telling the user when to trust the AI, without training, led to worse performance.
Importantly, the researchers’ system is fully automated, so it learns to create the onboarding process based on data from the human and AI performing a specific task. It can also adapt to different tasks, so it can be scaled up and used in many situations where humans and AI models work together, such as in social media content moderation, writing, and programming.
“So often, people are given these AI tools to use without any training to help them figure out when it is going to be helpful. That’s not what we do with nearly every other tool that people use—there is almost always some kind of tutorial that comes with it. But for AI, this seems to be missing. We are trying to tackle this problem from a methodological and behavioral perspective,” says Hussein Mozannar, a graduate student in the Social and Engineering Systems doctoral program within the Institute for Data, Systems, and Society (IDSS) and lead author of a paper about this training process.
The researchers envision that such onboarding will be a crucial part of training for medical professionals.
“One could imagine, for example, that doctors making treatment decisions with the help of AI will first have to do training similar to what we propose. We may need to rethink everything from continuing medical education to the way clinical trials are designed,” says senior author David Sontag, a professor of EECS, a member of the MIT-IBM Watson AI Lab and the MIT Jameel Clinic, and the leader of the Clinical Machine Learning Group of the Computer Science and Artificial Intelligence Laboratory (CSAIL).
Mozannar, who is also a researcher with the Clinical Machine Learning Group, is joined on the paper by Jimin J. Lee, an undergraduate in electrical engineering and computer science; Dennis Wei, a senior research scientist at IBM Research; and Prasanna Sattigeri and Subhro Das, research staff members at the MIT-IBM Watson AI Lab. The paper is available on the arXiv preprint server and will be presented at the Conference on Neural Information Processing Systems.
Training that evolves
Existing onboarding methods for human-AI collaboration are often composed of training materials produced by human experts for specific use cases, making them difficult to scale up. Some related techniques rely on explanations, where the AI tells the user its confidence in each decision, but research has shown that explanations are rarely helpful, Mozannar says.
“The AI model’s capabilities are constantly evolving, so the use cases where the human could potentially benefit from it are growing over time. At the same time, the user’s perception of the model continues changing. So, we need a training procedure that also evolves over time,” he adds.
To accomplish this, their onboarding method is automatically learned from data. It is built from a dataset that contains many instances of a task, such as detecting the presence of a traffic light from a blurry image.
The system’s first step is to collect data on the human and AI performing this task. In this case, the human would try to predict, with the help of AI, whether blurry images contain traffic lights.
The system embeds these data points onto a latent space, which is a representation of data in which similar data points are closer together. It uses an algorithm to discover regions of this space where the human collaborates incorrectly with the AI. These regions capture instances where the human trusted the AI’s prediction but the prediction was wrong, and vice versa.
Perhaps the human mistakenly trusts the AI when images show a highway at night.
After discovering the regions, a second algorithm utilizes a large language model to describe each region as a rule, using natural language. The algorithm iteratively fine-tunes that rule by finding contrasting examples. It might describe this region as “ignore AI when it is a highway during the night.”
These rules are used to build training exercises. The onboarding system shows an example to the human, in this case a blurry highway scene at night, as well as the AI’s prediction, and asks the user if the image shows traffic lights. The user can answer yes, no, or use the AI’s prediction.
If the human is wrong, they are shown the correct answer and performance statistics for the human and AI on these instances of the task. The system does this for each region, and at the end of the training process, repeats the exercises the human got wrong.
“After that, the human has learned something about these regions that we hope they will take away in the future to make more accurate predictions,” Mozannar says.
Onboarding boosts accuracy
The researchers tested this system with users on two tasks—detecting traffic lights in blurry images and answering multiple choice questions from many domains (such as biology, philosophy, computer science, etc.).
They first showed users a card with information about the AI model, how it was trained, and a breakdown of its performance on broad categories. Users were split into five groups: Some were only shown the card, some went through the researchers’ onboarding procedure, some went through a baseline onboarding procedure, some went through the researchers’ onboarding procedure and were given recommendations of when they should or should not trust the AI, and others were only given the recommendations.
Only the researchers’ onboarding procedure without recommendations improved users’ accuracy significantly, boosting their performance on the traffic light prediction task by about 5 percent without slowing them down. However, onboarding was not as effective for the question-answering task. The researchers believe this is because the AI model, ChatGPT, provided explanations with each answer that convey whether it should be trusted.
But providing recommendations without onboarding had the opposite effect—users not only performed worse, they took more time to make predictions.
“When you only give someone recommendations, it seems like they get confused and don’t know what to do. It derails their process. People also don’t like being told what to do, so that is a factor as well,” Mozannar says.
Providing recommendations alone could harm the user if those recommendations are wrong, he adds. With onboarding, on the other hand, the biggest limitation is the amount of available data. If there aren’t enough data, the onboarding stage won’t be as effective, he says.
In the future, he and his collaborators want to conduct larger studies to evaluate the short- and long-term effects of onboarding. They also want to leverage unlabeled data for the onboarding process, and find methods to effectively reduce the number of regions without omitting important examples.
More information:
Hussein Mozannar et al, Effective Human-AI Teams via Learned Natural Language Rules and Onboarding, arXiv (2023). DOI: 10.48550/arxiv.2311.01007
This story is republished courtesy of MIT News (web.mit.edu/newsoffice/), a popular site that covers news about MIT research, innovation and teaching.
Citation:
Automated system teaches users when to collaborate with an AI assistant (2023, December 7)
retrieved 25 June 2024
from https://techxplore.com/news/2023-12-automated-users-collaborate-ai.html
This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no
part may be reproduced without the written permission. The content is provided for information purposes only.
Opposition Leader Peter Dutton has announced that he will go to the next election promising to build seven nuclear power stations. Dutton has promised the sites can be operational between 2035 and 2037, and will be built on retired or retiring coal stations.
However, Swinburne renewable energy expert Associate Professor Mehdi Seyedmahmoudian says that while the plan could have some merit, our energy system can already be transitioned without relying on nuclear power.
“Peter Dutton’s approach could potentially reduce costs and accelerate deployment by utilizing existing infrastructure. However, advancements in renewable energy sources such as solar, wind, and hydropower, combined with energy storage technologies, offer more sustainable and efficient alternatives,” said Associate Professor Seyedmahmoudian.
“Innovations in battery storage, hydrogen fuel cells, and hybrid energy systems are significantly enhancing the reliability and cost-effectiveness of renewable energy.”
Nuclear energy typically emits very little carbon dioxide, with Dutton’s promise linked to Australia’s goal of achieving net zero carbon emissions by 2050. However, Associate Professor Seyedmahmoudian says renewables are a much safer option.
“Smart grid technologies, community microgrids, and demand response management systems optimize energy distribution and consumption, facilitating the seamless integration of intermittent renewable sources. This reduces the need for large-scale, centralized power plants and addresses environmental and safety concerns associated with nuclear energy, such as radioactive waste management and potential catastrophic failures.”
The price tag for the opposition’s nuclear promise is unknown, with Dutton confirming “comprehensive site studies” would be needed before a cost could be revealed.
Associate Professor Seyedmahmoudian says renewable energy is the answer for a cheaper and more secure energy future.
“By investing in research, innovation, and infrastructure for renewable energy and smart grid technologies, we can achieve a reliable, sustainable, and cost-effective energy transition without relying on nuclear power, ensuring long-term energy security and environmental sustainability.”
Citation:
Nuclear power has merits, but investing in renewable ensures long-term energy security, renewable energy expert says (2024, June 19)
retrieved 25 June 2024
from https://techxplore.com/news/2024-06-nuclear-power-merits-investing-renewable.html
This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no
part may be reproduced without the written permission. The content is provided for information purposes only.
