Lecture 10 - Misinformation.pdf

IT and Society Lecture 10: Misleading Information and Dark Patterns Prof. Jens Grossklags, Ph.D. Professorship of Cyber Trust Department of Computer Science School of Computation, Information and Technology Technical University of Munich June 24, 2024 Recap 1. Artificial Intelligence research, pr...

IT and Society Lecture 10: Misleading Information and Dark Patterns Prof. Jens Grossklags, Ph.D. Professorship of Cyber Trust Department of Computer Science School of Computation, Information and Technology Technical University of Munich June 24, 2024 Recap 1. Artificial Intelligence research, practice and deployment has a long and rich history with periods of excitement and despair. 2. Very different directions: E.g., modeling human intelligence versus creating actionable systems delivering output that is difficult to accomplish for humans. 3. Examples of AI for Good 2 Post-truth and Alternative Facts 3 Post-truth Relating to or denoting circumstances in which objective facts are less influential in shaping public opinion than appeals to emotion and personal belief Characteristics: – “Alternative facts” replace actual facts – Feelings have more weight than evidence – Experts seem to have lost authority 4 Question: Where does the widely shared aversion against numbers, science and technology come from? [The Big Lebowski] 5 Philosophical Perspective 6 Objective Truth? Perspectism. Friedrich Nietzsche (1844-1900): Objective truth is a relic of an age when God was the guarantor of what counted as the objective view of the world, but without God, objective, absolute truth is an impossibility Only alternative is a position called “perspectivism” – the idea there is no one objective way the world is, only perspectives on what the world is like We agree on things not because propositions are “objectively true”, but by virtue of sharing the same perspective http://theconversation.com/the-post-truth-era-of-trump-is-just-what-nietzsche-predicted-69093 7 Persuading people? At what cost? Funding goal: €1.5Million++ 8 Power Creates Dominant Views Michel Foucault (1926 – 1984) There is no absolute truth, just different regimes of truth. What you think can never be objective, it depends on your ideology (doxa) and the dominant discourse in your environment (episteme). There will always be dominating discourse that is the current truth. This discourse needs to be analyzed to find out who benefits and who is the victim (power relationships). This battle is continuous: there is no ‘best’ outcome. Everything depends on power relationships. (as cited by Beeckman) 9 Harry Frankfurt: Bullshitting vs. Lies “It is impossible for someone to lie unless he thinks he knows the truth … Producing bullshit requires no such conviction. A person who lies is thereby responding to the truth, and he is to that extent respectful of it. When an honest man speaks, he says only what he believes to be true; and for the liar, it is correspondingly indispensable that he considers his statements to be false. For the bullshitter, however, all bets are off … He does not reject the authority of the truth, as the liar does, and oppose himself to it. He pays no attention to it at all. By virtue of this, bullshit is a greater enemy of truth than lies are.” Do you agree? 10 Harry Frankfurt: Bullshitting vs. Lies Lack of knowledge breeds bullshitting: “Bullshit is unavoidable whenever circumstances require someone to talk without knowing what he is talking about. Thus the production of bullshit is stimulated whenever a person’s obligations or opportunities to speak about some topic are more excessive than his knowledge of the facts that are relevant to that topic.” 11 Research Perspective 12 How do humans decide? The Dual Processing Model Kahneman & Tversky 13 What does Behavioral Science Contribute? Some factors driving human perception: – Simplicity: People are much more willing to accept ideas they readily understand  Easier explanations will therefore win out – Confirmation Bias: People are much more willing to accept evidence in line with what they believe – Endowment Effect: People attach emotional value to their beliefs  This makes it difficult to change views/minds 14 Experiment by Gilbert et al. (1993) Subjects were exposed to false information about a criminal defendant (Experiments 1 and 2) or a college student (Experiment 3). Some subjects were exposed to this information while under load (Experiments 1 and 2) or time pressure (Experiment 3). Subjects made judgments about the target (sentencing decisions or liking judgments) Subjects were shown 30 sentences and were asked to classify each as (a) true statement from the first crime report, (b) false statement from the first crime report, or (c) statement that never appeared in the first crime report 15 Experiment by Gilbert et al. (1993) Both load and time pressure caused subjects to believe the false information to be truthful and to use it in making consequential decisions about the target. Our perception can be influenced; it is malleable 16 Recent Example (2023): Steered by Generative AI Language-model-powered writing assistant configured to argue that social media is good or bad for society Task: – Participants (N=1,506) were asked to write a post discussing whether social media is good for society – Participants also completed a social media attitude survey after the writing task Research Paper: Co-Writing with Opinionated Language Models Affects Users’ Views 17  Recent Example (2) The participant in the screenshot wrote an argument critical of social media, but the model is configured to argue that social media good for society. Research Paper: Co-Writing with Opinionated Language Models Affects Users’ Views 18 Recent Example (3): Written Text Research Paper: Co-Writing with Opinionated Language Models Affects Users’ Views 19 Recent Example (4): Attitudes Brackets indicate significant opinion differences at the **p