Monday, 29 April 2019

Week 4 [06-12.05.2019] The world's first online variantographer

A team from the University of Florida has developed a special algorithm that can be described as the world's first Internet lie detector. An online game was then designed in which players were randomly assigned to the role of "saint" or "sinner". The experiment involved 40 people who participated in 80 games. An intelligent algorithm analyzed the entries of people from both groups and distinguished specific words that can decide whether someone is a liar or not.

https://news.fsu.edu/news/science-technology/2019/03/15/no-lie-fsu-researcher-developing-worlds-first-online-polygraph/

Questions:
Is it possible to detect lies on the Internet?
Do you think that there will be an algorithm to search for false messages in online text?
Do you think that social informatics should focus on detecting fake internet content?
Do you think users should be informed that their messages will be scanned by this kind of lagorithm?

18 comments:

  1. Is it possible to detect lies on the Internet?

    In my opinion, you can detect lies on the Internet. As the author of the article recalls, the method of detecting lies is directed to social media such as facebook, twitter, etc. Most often in social media we speak the truth, because we are aware that our comments will be quickly verified by other Internet users. Therefore, you can use the profile information from the media as a comparative sample to our other comments on the same or similar topic. For the most part, the liars also use the description of things, people, phenomena, octants, they lie. They want to gain time to put together the most credible story.

    Do you think that there will be an algorithm to search for false messages in online text?

    It seems to me that it is rather. Because, according to the article, surprising research has already been carried out. You can also use chatbot, which will ask you the right questions, based on which you will be able to verify the statements. I think that in a few years such an algorithm will be widely available in order to improve the detection of lies.

    Do you think that social informatics should focus on detecting fake internet content?

    I believe that social informatics should help in verifying the content we receive. The reason is the excess of information that arises daily. The information generated on the internet should be verified. Obviously, people do the best at the moment, but due to the magnitude of the news, they are not able to verify everything. You need algorithms that can verify and mark reliable and unreliable information.

    Do you think users should be informed that their messages will be scanned by this kind of lagorithm?

    In my opinion, in most cases, users are aware that their comments / messages are verified by recipients. However, users should be informed about such proceedings. If you do not look it is their statement and they should bear responsibility for it.

    ReplyDelete
    Replies
    1. Thank you for answering. However, it seems to me that this information about data processing (GPEN) already exists.

      Delete
  2. Is it possible to detect lies on the Internet?
    It depend how much information we have about the topic. If the topic is well known and we have enough reliable data then yes it is possible. Someatimes as article says, the lie may be distinguish by some linguistic features. Ususally we do not know if the text is a lie if it is short and written about some new and uknown topic.

    Do you think that there will be an algorithm to search for false messages in online text?
    Yes I think someone is already working on such kind of algorithm. It may detect the truth or lies based in a text messages history or based on a messages to other people. However
    this kind of algorithm may disrupt privacy and that may be a problem for application of it.


    Do you think that social informatics should focus on detecting fake internet content?

    I think that social informatic is a wide discipline and there is a plenty of possibilietes for a topic. One of them is detecting fake internet content. I think it is importnat topic.
    Do you think users should be informed that their messages will be scanned by this kind of lagorithm?
    Yes the user should be informed as it interrupt privacy. The best option is when the user has to make special agreement for being scanned.

    ReplyDelete
  3. 1. Is it possible to detect lies on the Internet?

    Good question. I believe that only from the context of the statement we can infer about the truthfulness or lies of another person. Rather, the algorithm will not do it. I guess I may be wrong, still as people we set the decision criteria - despite neural networks.

    2. Do you think that there will be an algorithm to search for false messages in online text?

    Yes, on some beautiful day such an algorithm will rather emerge. Only what is he supposed to detect? Deceived words, or sentences, or analyze the context of speech? False messages. This is not my hand! Is it a true or false statement?

    3. Do you think that social informatics should focus on detecting fake internet content?

    Yes, social informatics should deal with such tasks. I think that's why it was singled out for her.

    4. Do you think users should be informed that their messages will be scanned by this kind of lagorithm?

    I will answer with a question. For what? Man, by definition, should be truthful. So why tell him that his statement will be analyzed. What could this mean for politicians?

    ReplyDelete
    Replies
    1. Then we must ask ourselves whether we have a right to silence or a right to lie.

      Delete
  4. Is it possible to detect lies on the Internet?

    It all depends on the case that will be taken into account. Some comments or accounts can be recognized with the naked eye and there is no need to use technology. But of course, depending on the advancement of the technology used to spread false information, it can be very difficult to track down.

    Do you think that there will be an algorithm to search for false messages in online text?

    I think, of course, not one such algorithm will arise and it is a very good idea. Of course, in any algorithm you can find gaps, or take advantage of its weak points and write a good enough bot that the algorithm will not recognize it.

    Do you think that social informatics should focus on detecting fake internet content?

    I am afraid that social informatics is full of people who have no clue about information technology from a technical point of view. Seeing even after our university, they study people after psychology, sociology and other thoughtless courses. I think that we should build a strong team of people who know about technologies and connect them with experts from other fields.

    Do you think users should be informed that their messages will be scanned by this kind of agorithm?

    I think so, because many of such information does not come from bots only from real people who have their goal in this activity. Perhaps this will reduce the amount of such information and fewer people will want to do so.

    ReplyDelete
  5. Hello Andrzej, thank you for an interesting article.

    To be honest, I was a little bit sceptical about the linked news because it was entitled to the word "polygraph" to exhibit its effectiveness. This word describes hardware that is intended to work in a specific environment and which is based on emotional reaction analysis on a given answer rather than wording.

    A presented neural network, fortunately, follows that approach because, to cite: "Researchers also calculated time lags between every sentence, and even parts of a sentence, by placing time stamps on the words. That precise breakdown clearly showed how much a person paused during interactions — another language-action cue.". So it shows that the researchers used not only dictionary under consideration, but also behavioural correlates that are combined with, for example, confidence.

    I think that it is a good approach and it could be used for lies detection on the internet. Application of that tool are very wide and they could be defined in scope that you've mentioned: from false messages detection in online texts to enterprise plugins to communicators (just to check if your business partner tells you a truth). I think it would be fair to inform the sender that his/her text would be scanned (just like in anti-plagiarism approach), however, I can see the situation when only one side knows about that fact.

    ReplyDelete
  6. The article provides some insight on linguistic cues that could be used to predict if a given text is truthful or not. It sounds reasonable, I think that at least to some extend it's possible to detect lies on the internet, especially if one had access to not only a text but also some additional data like facial expressions or voice.

    I think that any work towards detection of fake news would be valuable. I found a website for online content validation with AI: http://www.fakenewsai.com/ I've tested it on a few examples and it looks promising.

    Regarding informing the other parties that they're being analysed by some tool for lies detection, I don't think it's necessary. After all, if we engage in some communication we automatically agree that our listener judges what we're saying. If we have bad intentions and someone discovers it using any unharmful means it would be hard to blame him for doing so. In practice, informing someone about scanning their messages for detecting lies could deteriorate the possibility to do it accurately as the person would be more careful.

    ReplyDelete
  7. 1. Is it possible to detect lies on the Internet?

    I think it's possible to some extent. But as long as these solutions are created, people will try to avoid them somehow.

    2. Do you think that there will be an algorithm to search for false messages in online text?

    There are attempts to create such algorithms all the time. Nowadays, a lot of hope is placed in artificial intelligence algorithms. I think we are able to create such solutions that will detect obvious false information, but the more complicated the text, the more difficult it will be.

    3. Do you think that social informatics should focus on detecting fake internet content?

    I believe that this is one of the most important topics that should be dealt with in this area nowadays.

    4. Do you think users should be informed that their messages will be scanned by this kind of algorithm?

    Of course, such information should be public. Although the regulations are usually not read by users anyway. I think that asking many users many of them will not be aware that for example google analyzes the content of their emails.

    ReplyDelete
  8. 1. I think it is possible. If we have enough data about a topic from different sources. Especially it would be easy to detect lies in the media or newspapers. However, the detection of lies regarding, for example, someone's personal life (not celebrity) could be more difficult, because the amount of information on that person could be too small, or would put on the Internet only false entries that could not be verified differently than doing reaserch in reality.

    2. Yes. We already are able to use the algorithms to extract useful information from the message, eg customer satisfaction with shopping, so why not check the veracity of these texts.

    3. I think that it would be a huge help in posting only reliable information on the Internet. Many people are able to post fake entries to mislead others. Detection of false content would also be useful in the context of false medical information in the era in which a lot of people want to undermine the effectiveness of the latest research and medical results, saying that everything that scientists do is to harm us.

    4. It depends. Most people who lie on the Internet are aware of this and do so deliberately. Information about the operation of such an algorithm could make some people dissuade from posting false information, but some of these people would not care. What's worse could lead to a situation where especially such people post a lot of false posts to try to spoil the operation of the algorithm.

    ReplyDelete
  9. 1. Is it possible to detect lies on the Internet?
    I think it’s difficult to say who is a liar. When we talk with someone we could observe this person and mayby we will find some symptoms that tell us that this person is a liar. On the Internet when we don't see a person and only a message is displayed it could be difficult to decide if this is truth what the person wrote. The article show us that probably we can detect the liar. It’s very interesting and we will see what comes in the nearest future.

    2. Do you think that there will be an algorithm to search for false messages in online text?
    I think that probably yes. On the Internet there are many fake news. It’s difficult to decide what is true. This problem is know researchers and they work to find solution to help people filter fake messages. This article is a good example.

    3. Do you think that social informatics should focus on detecting fake internet content?
    I think yes. In the nearest future it could be a big problem. Today we can find web sites with fake news. I think this problem will be much bigger in the future. Social informatics should work to find a good solution to help fight with this messages and filter it from the true messages.

    4. Do you think users should be informed that their messages will be scanned by this kind of algorithm?
    I think that they should be informed about that. This can prevent people to write fake news and mayby in the future cut off this people from the Internet. I know it’s very aggressive method but in some situations it’s only option to protect web form this messages.

    ReplyDelete
  10. 1. It seems that - to some degree - it is possible. There might be some problems with precise timing information (as it was one of variables analyzed in the article) - if such algorithms gain wider adoption, we will have chat apps with not only end-to-end encryption, but also a jitter applied to our typing tempo to fool such algorithms.

    2. Yes, I think there will be. But those are two different things: message might be a truth (from the point of view of its author) and be factually wrong, or might be a lie while being factually true.
    Big scale research on relations between subjective and objective truth might be interesting thing.

    3. Once we recognize social informatics as important branch of science, then yes, it should be one of important goals of it. But, sometimes, a thought comes to my mind: can't we just end with this social media madness and get back to real life? And problem of fake internet content would solve itself.

    4. No, I don't think so. That would put too much power to the authors of messages, and at the same time limit rights of all other people. And, if such notification would be implemented, it would be just another pop-up to dismiss, just after cookies information (one of 10 dumbest things ever: why page publisher informs me that *my* browser running on *my* computer will save a cookie?) and GDPR notification.

    ReplyDelete
  11. Is it possible to detect lies on the Internet?
    Without the contextual clues that can hint at deception in a face-to-face conversation, computers will have problem with separating truthful people from liars
    Do you think that there will be an algorithm to search for false messages in online text?
    People already start working on those algorithms. The creators of a revolutionary AI system OpenAI, an nonprofit research company backed by Elon Musk that can write news stories and works of fiction – dubbed “deepfakes for text” – have taken the unusual step of not releasing their research publicly, for fear of potential misuse.
    Do you think that social informatics should focus on detecting fake internet content?
    Yes, fake news detection on social media is a newly emerging research area. The quick spread of fake news has the potential for calamitous impacts on individuals and society.
    Do you think users should be informed that their messages will be scanned by this kind of lagorithm? When we're dealing with humans, we have to be extra careful, especially when the implications of whether someone’s lying could lead to conviction, censorship.

    ReplyDelete
  12. 1. Is it possible to detect lies on the Internet?

    Yes, I am sure that it is possible to detect lies on the Internet. The easiest way to do that is verification the content by users and decide, if what they read is true or false. Unfortunately because of the existence of fake accounts on social media and other platforms, even such verification may not be correct, because persons standing behind such accounts can publish also incorrect and untrue statements or information, as well as opinions, which main goal is to support their (or somebody who paid them for such behaviour) point of view.

    2. Do you think that there will be an algorithm to search for false messages in online text?

    In accordance of what previous commentators wrote, I think that such algorithms, dealing with the problem of searching for false messeges in online text, are invented and implemented all the time.

    3. Do you think that social informatics should focus on detecting fake internet content?

    From what we can conclude by analyzing latest conditions, fake news have a big impact on proccess of making decisions for entire society, also in cases important for the community. In my opinion detecting such untrue information and telling about them may have a positive influence for all of us and can cause the possibility to make decisions in the best way.


    4. Do you think users should be informed that their messages will be scanned by this kind of lagorithm?

    I have no doubt that users should be informed that their messeges will be scanned by this kind of algorithm. First of all, everyone should know that what he or she wrote may be analyzed by some third parties, not only by the addressee of the information. What is more, such knowledge may discourage some part of users to publish untrue information.

    ReplyDelete
  13. 1. Is it possible to detect lies on the Internet?
    If such well-known professor is conducting research in this direction, he probably does not waste his time. So the answer is obvious - it is possible. The only question is with what accuracy. The current results are impressive.

    2. Do you think that there will be an algorithm to search for false messages in online text?
    Yes, I think that such algorithm will be created. On the other hand, people also have a problem to properly assess the intentions of the other person, relying only on messages. Therefore, I am only concerned about the effectiveness of that algorithm. Ethical issues are not as obvious as it seems...

    3. Do you think that social informatics should focus on detecting fake internet content?
    As far as I’m concerned social informatics deals with fake Internet contents, for example the detection of fake news. Fake Internet content is a part of social life and if it appears on the Internet, it is a subject for scientific analysis.

    4. Do you think users should be informed that their messages will be scanned by this kind of algorithm?
    Good question. Since this is not personal data and is not distributed, the owner does not have to be informed about their processing. Similarly, if you write something and someone judges you on the basis of your published text, then you should be informed about it? I do not think so.

    ReplyDelete
  14. 1. Is it possible to detect lies on the Internet?
    I think, that theoretically it is possible, but it would require sophisticated algorithms of text analysis, which truly "understand" the meaning of the texts and can compare the information from various sources. Machine learning would be the solution, as in many cases.
    2. Do you think that there will be an algorithm to search for false messages in online text?

    To my mind, if a serious problem exists, if it poses threat and potentially can cause dangerous consequences, it will attract the attention of scientists and false messages are no exception, if they start causing serious problem, the researchers from various fields will focus closer at detecting them.

    3. Do you think that social informatics should focus on detecting fake internet content?
    I think that it will be the good solution if the social informatics are focused on the detecting fake internet content. Nowadays, there are a lot of fakes on the Internet, so, to my opinion, the issue is serious enough for scientists to work on it.

    4. Do you think users should be informed that their messages will be scanned by this kind of lagorithm?
    I think the users should be informed about that kind of scanning, it can possibly decrease the activity of the persons spreading fake news.

    ReplyDelete
  15. Hi,
    Honestly, I do not believe in lie detector even in its classic version. I was examined using this many years ago and while I was speaking true, the tool told “lie”. Maybe there are more precise algorithms now, but this is utopian idea to believe in the results of such experiments. In my opinion it looks even less optimistic with such virtual lie detectors, because apart from the risk of improper interpretation of human speech, we have to cope with analyzing potential intention of people – it is much easier to understand it in direct personal contact rather than through written messages, without someone’s physical presence.
    Summarizing, I do not believe it is possible to detect lies in the Internet.
    Consequently, there will never be trustworthy algorithm to search for false messages in online text.
    Detecting lies can be organized only with people, who do concentrate on this, reading the text with understanding, but they should also have very good psychological skills and knowledge of human’s mind. So, yes, social informatics should focus on detecting fake internet content. But informatics with not only technical knowledge and experience.
    In my opinion, there is no need to inform people about potential scanning their information. For me result of such examination is not trustworthy, so I’d rather consider this as opinion about the text given, and opinions are not forbidden.
    BR
    Marta

    ReplyDelete
  16. Achkar law in Toronto is the best law firm working
    for the peoples.
    They have
    Employment Lawyer Toronto for the peoples who do not know their rights and working in
    a company
    Many law firms

    ReplyDelete