Tech & Science Discrimination by algorithms - study proves cases and recommends countermeasures

17:10  25 november  2020
17:10  25 november  2020 Source:   t3n.de

What Moderna’s Covid-19 vaccine results mean for ending the pandemic

  What Moderna’s Covid-19 vaccine results mean for ending the pandemic Moderna says its Covid-19 vaccine is 94.5 percent effective. Here’s what you need to know.“This is a pivotal moment in the development of our COVID-19 vaccine candidate,” said Stéphane Bancel, CEO of Moderna, in a statement. “This positive interim analysis from our Phase 3 study has given us the first clinical validation that our vaccine can prevent COVID-19 disease, including severe disease.

"In this case , the algorithms generated will also do so. When processing data containing evaluations of people by other people, unequal treatments and discriminations may even Recommendations of Countermeasures . The study lists several options to counteract discrimination by algorithms .

Our study shows the generalization ability of one countermeasure over the other. This work presents rst analysis of spoong countermeasures for attack dependency and generalization. [12] P. L. De Leon, B. Stewart, and J. Yamagishi, “Synthetic speech discrimination using pitch pattern

For a study by the Bundestag, scientists took a closer look at the false assumptions of artificial intelligence. Some of these errors have been identified as discrimination. The technology follow-up office suggests solutions. Unter anderem kritisiert das Technikfolgen-Büro die Diskriminierung durch  Gesichtserkennungssysteme. © metamorworks / Shutterstock Among other things, the technology follow-up office criticizes the discrimination by facial recognition systems.

"Social inequality of treatment can be inscribed in complex algorithmic decision-making systems (AES) and thus potentially affect a large number of people", write the authors of the study "Possible discrimination through AES and machine learning" . After general explanations, the Office for Technology Assessment at the German Bundestag (TAB) proves its thesis using four striking case studies. The TAB consists of experts from various institutes in this field under the leadership of the Karlsruhe Institute of Technology.

After the authors have divided algorithmic systems into self-learning or rule-based, they bring AES into play as a higher-level entity. They carry out the data acquisition and analysis, interpret the results and derive recommendations for decisions. They are not only used as preparation and support, but also as a substitute for human decision-making processes. The study gives examples of such systems and problems.

How shifting from meat-heavy to plant-based diets can help reverse the climate crisis

  How shifting from meat-heavy to plant-based diets can help reverse the climate crisis A new study shows that moving to a plant-based diet is critical, but governments have been slow to act.These fires are but the latest distress signal from a deeply unsustainable global food system. Emissions are embedded in every part of the food supply chain, from deforestation to grow crops or raise cattle, as in Brazil — which releases carbon dioxide and nitrous oxide — to rice fields and cow burps, which emit methane, another potent greenhouse gas.

Countermeasures on the present situation and prevention of financial risk management in construction enterprise. Although there have been proven and successful developments in the field of Higher Education for Sustainable Development over the past 15 years or so, there are still numerous

Microeconomics self- study problems. Thus in addition to the SE that reduces leisure we observe income effect that makes that increases demand for leisure (in this case leisure must be a normal good) and it might happen that IE dominates the SE so that total demand for leisure increases

Discriminated by the algorithm

The first example is an AES in the medical field. The system calculated a lower death rate for patients with multiple illnesses and chronic illnesses than for those with pneumonia alone. As a result, it assigned a lower need for care to the seriously ill. The error lay in the data with which the system had been trained: in those, chronically and multiply ill patients had already received intensive treatment and therefore had a lower mortality rate.

Video: Infection Protection Act: The facts at a glance (glomex)

The second case comes from the Austrian employment office. The system divides clients into three categories, from close to the labor market. Women with care responsibilities are classified as remote from the labor market. The background: gender alone leads to a deduction in proximity to the labor market, and the duty of care to another. An overall view of the processes must show whether there is discrimination here, the authors say.

The Last Children of Down Syndrome

  The Last Children of Down Syndrome Prenatal testing is changing who gets born and who doesn’t. This is just the beginning.Every few weeks or so, Grete Fält-Hansen gets a call from a stranger asking a question for the first time: What is it like to raise a child with Down syndrome?

A case study is a report of an event, problem or activity. A case study format usually contains a hypothetical or real situation. An effective case study example should: Have their basis from real-life situations but you may conceal identities. Consist of a number of parts which would end with points for

Discrimination is the act of making unjustified distinctions between human beings based on the groups, classes, or other categories to which they are perceived to belong.

example three concerns the likelihood of recidivism in the United States. The Compas system used for this was wrong in predicting the quota for African American offenders. Forty-five percent of the group stayed law-abiding, while only 24 percent of the white candidates did.

Many AES in the USA that make up example four are subject to a structural error. Their facial recognition systems most often misidentify female and dark skinned faces. They are used in the penal system, law enforcement, internal security and prevention. Both false-negative and false-positive results can have "unpleasant consequences" for the individual, according to the 80-page document. IBM had ended its face recognition program for this reason .

Broad options for action suggested

The study emphasizes: The discrimination at the Austrian employment office would not have been noticed if the system had not been transparent. The scientists therefore plead for more transparency and control, for example through an independent body. In addition, mandatory labeling could help to signal the use of AES to those affected. However, it should not, as with common data protection declarations, lead to pass-through forms without an increase in knowledge. The study suggests documenting and logging data sets and models. Other ideas consist of establishing a competence center, an approval system and tests by anti-criminal bodies. In the area of ​​regulation, liability law should be adjusted, group discrimination should be included and collective legal protection should be installed, for example through collective actions. In addition, regulated self-regulation, for example through certificates, could help. The data ethics committee had recently made a similar statement and had exposed itself to all sorts of criticism from the digital associations .

The Bundestag's technology impact office proposes a variety of measures to curb discrimination through AI systems. (Screenshot: TAB / t3n)

The society should decide

In the end, the TAB establish a development towards an algorithmic society. It requires a reflection on which values ​​govern AES, what consequences they have and how they should be regulated. Social decisions need to be made at the development stage. The paper calls for them to be handed over to opinion-forming, democratic processes and legislative levels.

Why you shouldn't keep a tampon in for longer than 8 hours — and what to know if you do .
You shouldn't keep a tampon in for longer than 8 hours, as this can increase your risk of serious conditions like Toxic Shock Syndrome. Though this condition is rare, it's important to remove a tampon as soon as possible if it's in for more than 8 hours, and seek medical attention if you develop a high fever, rash, or diarrhoea. It's best not to sleep with a tampon in, and if you're swimming or sitting in a bath with a tampon in, you should change it once you get out. Visit Insider's Health Reference library for more advice.

usr: 3
This is interesting!