Minds and Machines PDF - Biases, Heuristics, and Decision Making

Minds and machines ✅ ✅ 1.​ Introduction ✅ 2.​ Biases and Heuristics ✅ 3.​ O...

Minds and machines ✅ ✅ 1.​ Introduction ✅ 2.​ Biases and Heuristics ✅ 3.​ Overconfidence ✅ 4.​ Bounded Awareness ✅ 5.​ Emotions (Zoom session) ✅ 6.​ (Emotions ) ✅ 7.​ Framing ✅ 8.​ Choices and Nudges ✅ 9.​ AI and moral decision ✅ 10.​ (guest lecture ) ✅ 11.​negotiations ✅ 12.​(guest consumer behavior ) 13.​Collab and group 14.​collab and decision strategy 2: Bias and Heuristics Recap Prescriptive VS Pre = Rational, how it “should” be Descriptive models Ex: Weighted additive rule: define prob, identify criteria, weigh criteria, generate alternatives, rate alternatives, compute optimal decision - assumes que have all the info all the time system 2 Desc = how it really is, with emotions, limitations, heuristics -​ system 1 reasoning Global vs bounded Global Rationality rationality -​ complete info, well organized preferences, excellent computational skills -​ Maximizing (seeking best solution, spend too much time and effort) -​ system 2 reasoning Bounded rationality - system 1 reasoning -​ limited info access, uncertainty, changing preferences, MAO varies, social context matters -​ satisficing (choosing good enough, spend less time and effort, miss out on optimal solution) → usamos Heuristic pra minimizar time and effort! → to make quicker judgments under uncertainty! -​ Examining fewer cues, Reducing the difficulty associated with retrieving and storing cue values, Simplifying the weighting principles for cues, Integrating less information, Examining fewer alternatives -​ we can overrun the influence of heuristics by using system 2 -​ can lead to bias! (systematic errors in judgment) Availability = estimating probability of events based on how easily you can think of examples of it - the ease of recalling make it heuristics seem more probable then it actually is Bias 1: Ease of recall -​ qto mais vivid ou recent or emotional um event, mais vc acha q é probable -​ ex death by starvation rate is way higher then by diseases. Airplane creases sao super improvaveis. Mas lembramos mto mais facil dessas coisas -​ power of story telling! -​ rare events and unique stories are easier to recall because they stand out e are memorable, msm que less frequent do que more normal events Bias 2: Retrievability -​ ease of retrieval from memory based on memory structure -​ aka o contexto que estamos faz ser mais facil de pensar em certas coisas -​ ex lugar de produtos no supermercado ⇒ The ease of recalling such events makes them seem more common than they are, leading to wrong probability assessment Representative- = This heuristic is used when ppl judge the probability of an event by comparing it to an existing prototype in their ness minds → assuming similar outcomes just because of similarity ⇒ leads to neglecting base rates and other relevant information heuristics Representativeness definitions: How similar an example is to members of its class How similar a sequence is to what we think randomness looks like ex: -> just bc Tom looks nerdy we think he is in Library science, although if we had to randomly guess someones degree we should say bwl q é tipo 30% dos alunos na TUM. → nossas predictions sao based em similarity, e nao em base rates como deveriam ser Bias 3: Base rate neglect -​ ignoring the general statistical information (base rates) in favor of specific information about a case. Judging statistics based on stereotypes/representativeness instead of base rates. Bias 4: Ignoring sample size -​ Failing to recognize that smaller samples have more variation - e asim assume that a small sample is representative of the whole -​ although smiles sample sizes have more variation, we still think the chances of getting 60% heads em 3000 flips é mais provável do que em three flips -​ ex: if a manager makes decisions based on feedback from a small focus group, they might ignore the potential for greater variability in opinions compared to a larger, more diverse sample Bias 5: Misconception of chance -​ Believing that random events should look random -​ "gambler's fallacy" = we believe that after a series of losses, a win is "due" to occur, despite each event being independent (ou usar o msm nr smp mas trocar dps de ganhar com ele pq agr a chance dele vir dnv é menor, mas isso é mentira) Bias 6: Regression to the mean -​ overlooking that extreme events are often followed by more average ones due to the natural tendency of data to regress to the mean -​ vc expect um really hot day after a really hot day, even though an average temperature day is more likely Bias 7: The conjunction Fallacy -​ Assuming that specific conditions are more probable than a single general one. -​ acontece when the conjunction appears more representative than the component descriptor -​ a chance dela ser A+B é obv menor, mas como a gnt acha que as coisas fazem sentido juntas ou que B representa mto, então achamos que a chance desse conjunction é alta Confirmation = tendency to search for, interpret, and remember information that confirms one's preconceptions heuristics Bias 8: Confirmation trap -​ seeking to confirm prior info - look for info that already matches what we think, looking for patterns - while ignoring contradictory evidence  Bias 9: Anchoring -​ Relying too heavily on an initial or unrelated piece of information - ppl tend to have an arbitrary starting point (ex last digit do seu cel pro ano do taj mahal) -​ own anchor (People develop estimates by starting with an initial anchor that is based on whatever information is provided) -​ External Anchor: The existence of an anchor leads people to think of information that is consistent with that anchor -​ Occurs even when anchors are presented subliminally - Ex mere exposure to high prices increases ppl willingness to pay, ou usar “before price” as anchors na etiqueta so you think you are making good deal Bias 10: Conjunctive and Disjunctive Events -​ Overestimate the likelihood of high probability events occurring consecutively -​ Underestimate the probability of a low probability event occurring once over many attempts -​ ex: em obras q tem multi-steps, ppl underestimate tudo que pode dar errado Bias 11: Hindsight and -​ Believing past events were more predictable than they actually were -​ “I-knew-it” bias - Knowledge of an event’s outcome works as an anchor Bias 12: the Curse of Knowledge -​ when assessing others’ knowledge, people are unable to ignore knowledge that they have that others do not have -​ Diversity as a remedy! adopt a mindset of perceiving differences, aks questions, feedback, visuals, etc Bias 13: Overconfidence… -​ This bias occurs when individuals overestimate their knowledge, abilities, or the accuracy of their predictions. It can lead to poor decision-making and excessive risk-taking Cognitive Bias = visual map that categorizes over 180 known heuristics and cognitive biases. Codex Categories relate to four problems: Too much information: information overload leads us to filter and make selections. Not enough meaning: we fill in gaps with stuff we already think we know, by making assumptions and patterning. Need to act fast: we’re constrained by time and information, so we simplify and use shortcuts. What should we remember:We need to make trade-offs around what we try to remember and what we forget. 3: Overconfidence Overconfidence definition = Individuals overestimate their knowledge, abilities, or the accuracy of their predictions. -​ can lead to poor decision-making and excessive risk-taking -​ super robust and well studied an facilitates other biases -​ consequences: being in debt, dotcom bubble, going to war, titanic, chernobyl Three forms/manifestations of overconfidence: types of overconfidence Overprecision = Being too certain that we know the truth/ about the accuracy of our beliefs Manifestations: -​ excessive certainty in our accuracy -​ Lack of interest in testing assumptions -​ dismissing evidence suggesting we are incorrect -​ overly narrower confidence intervals! Causes -​ Lacking knowledge - experts are more precise BUT overconfidence persists and draw narrower intervals -​ we are rewarded for overprecision - quem é viewed as confident are seen as better/more capable e more  trusted! -​ we don't seek contradictory evidence - confirmation bias Consequences -​ ignoring feedback and alternative viewpoints (em vez de average them) -​ underestimating risks and uncertainties → overly optimistic plans and potential costly errors -​ inaccurate predictions with too narrow confidence intervals → poor planning -​ reduced collaboration because of too rigid thinking -​ Inefficient allocation of time, money, and other resources Mitigation -​ averaging more estimates + consider alternative outcomes Overestimation = Overestimating one's actual abilities or performance (overly optimistic) Manifestations: -​ overestimating our abilities, control, success, knowledge, speed to complete tasks example -​ Entrepreneurs and managers often overestimate the success of their ventures or the time it will take them to complete certain task Consequences: ⇒ inflated sense of one's capabilities or the likelihood of positive outcomes: lead to unrealistic expectations and poor decision-making bc we don't adequately prepare for the real scenario! Resulting on delays or failures of projects Self-enhancement -​ ppl rather see themselves positively as opposed to accurately (incl. groups we belong to) -​ effects are unconscious and strongest when responding fast/automatically (system 1!) -​ pode ser bom por evolutionary motive, tem sim q believe in yourself pra do stuff e mental health Optimistic biases -​ Unrealistic optimism: The tendency to overestimate the rosiness of our future -​ Defensive Pessimism: we brace ourselves for disappointment by making pessimistic assessments about our own abilities, status, and future performance -​ Loss aversion: Losses are more painful than gains -​ “Moment of truth” effect: We start off full of hope regarding an unknown future outcome, but as the moment of receiving actual performance feedback draws near, we tend to reduce our expectations The planning fallacy -​ tendency to overestimate the speed at which we will complete projects and tasks - especially when large -​ why: we forget to plan imprevistos e try to stick to plan-based scenarios -​ underestimation of our speed for simple tasks (acha q demora mais) -​ we overestimate our plans, but not those of other ppl -​ Even if we have relevant past experience, we tend to diminish its relevance by attributing it to external, transitory, or specific causes Overplacement = Believing one is better than others, often seen in competitive contexts Manifestations: -​ Generally, ppl think they are better than average! -​ Undeplacement can happen with difficult task or rare events -​ a gnt se avalia neglecting the base rate (tipo de vdd, qual a chance de alguém ser top 1% da sala, e nao qual a sua chance de ser 1%) Example: 90% of drivers rate themselves as better than average, which is statistically impossible. Consequences: (business contexts) -​ excessive competition and poor cooperation within teams -​ Inflated expectations of success, Excessive legal costs, Over-entry into markets (achando q todo mundo va querer o seu produto pq é melhor q o resto), M&A (acha q vc vai ser melhor liderando do q os outros), Choosing the wrong occupations. Dealing with overconfidence “Side effects” -​ positive illusions can have benefits for your mental health/ambition -​ porem, podem tb levar à inability to learn from mistakes (tenta externalizar tudo), put in less effort in tasks, hits to our credibility, dishonest behavior how to mitigate try to match private beliefs to reality (n how u wish things were) - can be difficult when objective data to larn from is missing. The key is: + Awareness + calibration (feedback from ppl with different perspectives and realistic assessments) + use of data (rely on objective data e not intuition, past experience) strategies: ⇒ draw wider confidence intervals, acknowledge that you might err! 1)​ become aware of biases 2)​ make data informed decisions (and use tools to keep better track of data) 3)​ plan multiple scenarios 4)​ make real time adjustments 5)​ get diverse feedback business context 1)​ Watch-out when relying on historical data: stable trends can quickly change. Incorporate external market factors like economic conditions and trends, and regularly review past forecasts to learn from errors. 2)​ Diverse data and perspectives: Digital signals (Google Trends, social media, customer sentiment), influencers and trend spotters, multiple diverse perspectives. 3)​ Multiple scenarios: Best-case, worst-case, and most-likely scenarios to manage volatility of market and trends. Pilot tests and real consumer feedback to validate assumptions. 4)​ Monitoring and adjustments: Continuously monitor market changes and adjust forecasts. Use contingency planning to prepare for risks and seize opportunities or respond to shifts quickly. Consider an early warning system. 4: Bounded Awareness Overview definition = mental limitations that prevent people from noticing, seeking, or using available and relevant information. This can lead to focusing on less important data while ignoring crucial information -​ Another Bias Resulting from Effort Reduction -​ acontece msm sem information overload, mas internet hj em dia deixa muito pior Características: -​ Limited attention, misdirected attention (coisas que nao são tao importantes), can only see one thing at a time, misses unexpected changes in the environment BA in decisions: in various points of the decision making process -​ fail to see or seek out info needed to make decision -​ fail to use info we do see bc we aren't aware of its relevance -​ fail to share info with others e assim bounding others’ awareness Ex: -​ kodak e blockbuster não deram atenção ao que era importante e morreram linked to bounded Lembrando que bounded rationality é mais sobre limited info access, nao conseguimos pensar sobre todas as rationality opcoes, nao temos todas as infos, changing preferences, social context matters mitigation individuals and groups should: -​ actively seek out unshared information -​ broaden their attention Inattentional Blindness definition = fail to notice unexpected stimuli, even when looking directly at it just bc you weren't looking for it Ex: -​ video do gorila ou piloto nao vendo outro aviao pq tava focused no seus controls -​ na alfandega os que tavam expecting bad objects de fato encontraram eles mais facil! Consequences: -​ overlooking a broad array of info that is already available! -​ creativity problems → bounds can eliminate good solutions. Se vc tentar sair dessas bounds e das suas prior assumptions vc pode enxergar outras respostas! Think outside of the box! -​ Product developers should try to unbound them from their assumptions by imagining new products as if resources werent a constraint Functional = specific type of bounded awareness - looking at tools only in the way they have always been used Fixedness -​ to avoid it try to think outside of the box and have cognitive flexibility! New options will appear! -​ ex: Netflix from DVD rentals to streaming Change Blindness definition = ppl often fail to see changes, especially when it is unexpected -​ e se for introduces during eye movement e for gradual -​ Attention is needed to see change -​ only a single change can be seen at any moment pq tem q process it cognitively -​ slippery slope of unethical behavior - pq we nao notice little changes, então vai de pouco em pouco Focalism Focusing illusion = tendency to make judgments based on attention to only a subset of available information, overweighting that information and underweighting unattended information. It can lead to skewed perceptions and decisions. - Focusing on a single event demais e assim ignorar outras coisas importantes → data visualization influencia isso! Pq te mostra em que info tem que focar Affective = tendency for people to inaccurately predict their future emotional responses to events (incl. intensity and Forecasting Errors duration of emotion) -​ ex ganhar na loteria/new job acha q vai ser feliz assim pra sempre -​ avoid focusing on the data that you have, and try to think about data that wold best answer the question Status quo bias = tendency to prefer current conditions (status quo) or the existing state, even when alternatives may offer better outcomes, como estar bounded by current knowledge and comfort zones you miss new info that changes status quo Impact of status quo bias: -​ Resistance to change -​ Overlooking emerging risks and opportunities (blockbuster) -​ only have a narrow scope of alternatives mitigating focalism 1)​ Be aware of where your attention is directed to: data visualization matters and guides us to specific problem aspects 2)​ Diversify attention and unpack a problem: Consider all aspects of a problem or decision, not just the most prominent. 3)​ consider long-term and multiple scenarios of a choice - assim pensa em coisas diferentes, n foca 4)​ Seek broader and diverse feedback Bounded awareness in groups and strategy settings groups -​ groups tem combined information! -> more comprehensive decision making -​ porém, groups discussions ficam presas no “mentioned information” effect: normally focus apenas on shared information only, e nao new individual e unique info -​ group awareness is constrained by what is discussed -​ Open source movement: encourage info sharing de unique experiences and collective innovation strategic settings -​ bounded awareness can prevent individuals from fully understanding the implications of their decisions or the behavior of others -​ ex in negotiations 5: Emotions (guest lecture) texto that the most effective way to maximize customer value is to move beyond mere customer satisfaction and connect with customers at an emotional level That means appealing to any of dozens of "emotional motivators" such as a desire to feel a sense of belonging, to succeed in life, or to feel secure, stand out form crowd, bring order and structure, belonging, thrilled by the experience, freedom and independence Through emotional connection = powerful way to increase costumer value drives significant improvements in financial outcomes - more than twice as valuable as highly satisfied customers. These emotionally connected customers buy more of your products and services, visit you more often, exhibit less price sensitivity, pay more attention to your communications, follow your advice, and recommend you more Pode fazer na marketing campaign, pode fazer na hora que abre o site, mandar personalized notes, Idea is to implement an emotional connection based strategy across the entire customer experience porem, these connections are unconscious! costumers cant really explicitly say that this is an important aspect in the costumer journey, even though it does have a hug impact 6: Emotions in decision making II Emotion vs. cognition Role of emotions -​ emotions are also important in decision making - nem todas nossas decisões são racionais assim What are emotions = brief episodes of synchronized responses that produce noticeable changes in an organism. They involve physiological arousal, motor expression, and subjective feelings (butterflies in stomach). Brought up by triggering events. They are also protective: emotion as evolved capacity 1)​ Disgust: repulsive response to purge from contaminants to be safe -​ Implication for consumer behavior: esse feeling nos faz querer get rid of products, enato sellers abaixam preço pra gnt considerar comprar ainda 2)​ Fear - can make us risk averse, super primitive, kinda survival 3)​ surprise & negative info - such emotions make you more interested (directs focus) thats why fake news spreads emotion VS. internal conflict between different parts of an individual's self, often described as the: cognition: -​ "want self": following emotions, mais immediatists two selves -​ "should self: analytical thinks about payoff (bolo saudavel) -​ Decision utility vs experience utility: temos um hard time predicting feelings (and their intensity) e utility de actions! aqui tb tem um disconnect! Hyperbolic discounting (impact of temporal differences) -​ ppl normally tend to chose quick small gratifications instead of larger delayed rewards como resolver: -​ usar commitment devices (schedules, automatic savings), Communication between the two selves (planning and reflection) and compromise (negotiate between two selves ⇒ Understanding these temporal dynamics is crucial for making decisions that align with long-term objectives and avoiding the pitfalls of short-term thinking The influence of emotion on cognition/ decisions affect and decision how emotions influence judgments: making 1)​ Risk and benefit judgments - emotions affect como vc perceive risks e benefits, ex fear can lead to more risk-averse decisions 2)​ Evaluability Principle - is the ease with which we can evaluate an option affects our decision-making. Emotions influence this evaluability, making certain aspects of a decision more salient. 3)​ Emotional attachment - can drive decisions aka consumer behavior and brand loyalty basic emotional how do emotions influence our decisions?: processes 1) emotions direct Emotions can Direct Attention: attention -​ emotions focus our attention on specific stimuli -​ which influences information search -​ Positive emotions (e.g., satisfaction, joy, pride) can stop the search; while anxiety (medo) makes you search for more alternatives; and anger makes you search for aggressive alternatives to achieve revenge 2) emotions can (MOA) affect depth of -​ signal when a situation demands more attention or when heuristic processing is sufficient thought -​ Negative mood = signals threat, increave vigilance, system 2 -​ Positive mood = signals safe environment, back to default, system 1 3) appraisal -​ jeito mais atual de look at emotions tendency framework -​ emotions dao appraisals (evaluations) de situações que nos ajudam a fazer decisões. -​ predictable patterns in decision making: -​ Anger: Increases blame attribution, often leads to punitive decisions. Fear: Leads to risk-averse choices, focusing on safety and caution. Happiness: Associated with optimism, potentially increasing risk-taking. 4) Integral = emotions that arise directly from the decision at hand. emotions -​ directly linked to the outcomes or consequences of the decision. How they affect decisions: como um gut feeling! vc se guia por isso tb! -​ can be a source of bias (pq emotionally intensive events lead to availability heuristic) -​ can help making adaptive decisions! vc aprende como agir melhor da próxima vez, especially in time-sensitive or high-stakes situations -​ intense emotions attached to terrorist attacks may make those events easier to retrieve from memory - pessoas preferiram drive instead of flying a while after 9/11 5) Incidental = are present at the time of decision-making but are unrelated to the decision itself. They can influence decisions emotions even though they are not normatively relevant. -​ ex os hungry judges - na vdd huger n tem nada a ver com dar a sentence, mas afeta ainda assim How they affect decisions: feelings que influenciam sua percepção, mas n causadas pela situacao -​ sao emotions de outras situations que são carried over to essa situation, mas ainda influence the decision -​ can lead to unintended bias 6) Regret avoidance = making decisions to minimize the possibility of future regret -​ ex: comprar o quarto de hotel agr na promoção antes q acabe -​ stick with a default/first option or avoid making changes to avoid the regret of a potentially worse outcome Recognizing 1)​ Awareness reflection Emotions and -​ recognize current emotions before making a decision (mindfulness or pause) Overcoming 2)​ Reappraisal techniques Biases -​ reframe the emotional context by interpreting events in a different, less emotional light 3)​ implementing time delay -​ allow time between the emotional event and decision making to reduce impulsivity How machines recognize emotions Applications for testing new ads and products, coding of emotions of customers on social media, customer satisfaction in stores, emotion recognition real time customer service, educational tools to track engagement/confusion, gaming to adapt emersion.. 1) Physiological = Monitor physical indicators (e.g., heart rate, skin conductivity) Data -​ wearable fitness trackers medical applications 2) Facial recognition = Analyze facial movements and expressions, mapping them to specific emotions. -​ detect facial landmarks and track micro-expressions. Convolutional neural networks (CNNs) and facial action coding systems (FACS) help interpret emotions from facial data. 3) Audio = through vocal characteristics such as pitch, tone, volume, and speed -​ speech signal processing and neural networks trained on labeled emotional speech datasets -​ ex: call centers e virtual assistants that analyze voice to find emotion and adapt in real time, 4) text = emotional cues using sentiment analysis, keyword spotting, and deep learning -​ ex: chatbots that detect emotion real time, Social media sentiment analysis to gauge public reactions to events, products,or campaigns -​ Challenges: -​ ambiguity and nuance (Irony, sarcasm, deception, abbreviations, emojis) -​ context (ex: great disappointment) -​ deliberate contrasts Business application: -​ monitor social media to understand brand perceptions over time -​ Analyze product reviews to identify portfolio innovation opportunities -​ Prioritize and address customer complaints to increase customer retention social emotions Group decision -​ emotions spread through groups and create “shared moods” that impact decisions making and -​ can amplify risk taking or caution from the group emotions -​ foster groupthink, and not disagreement -​ Group decision outcomes are influenced by emotional alignment or conflict Emotions and -​ Certain emotions, like disgust, heighten moral sensitivity and judgmental attitudes - ex puxar a Ethical Judgements alavanca e mudar o curso do trem -​ Anger and indignation can lead to harsher ethical evaluations -​ Emotions may override rational ethical frameworks, leading to inconsistent judgments (hunger) -​ Ethical decisions are prone to bias from incidental emotions -​ Empathy towards machine changes the way we handle technology? 7: Framing: how info presentation impacts decisions Are we too impatient to be intelligent? Text The text explores the tension between our desire for immediate gratification and the need for thoughtful, intelligent decision-making Impatience and Decision-Making: People often prioritize short-term rewards over long-term benefits, leading to impulsive decisions that may not align with their best interests. This impatience can undermine intelligent decision-making, as it discourages thorough analysis and consideration of future consequences. Cognitive Biases:  The text highlights how cognitive biases, such as overconfidence and optimism, can exacerbate impatience by making individuals believe they can achieve quick success without careful planning. These biases can lead to overestimating one's abilities and underestimating risks, resulting in poor decision outcomes. Balancing Short-Term and Long-Term Goals: The challenge lies in balancing the immediate desires with long-term objectives. Strategies like setting clear goals, using commitment devices, and fostering self-awareness can help manage this balance. Role of Technology and AI: The text also discusses how digital environments and AI can both amplify and mitigate these biases. While technology can provide quick solutions, it can also offer tools for better decision-making by providing data-driven insights. Overall, the text emphasizes the importance of patience and strategic thinking in making intelligent decisions, especially in a fast-paced, technology-driven world. Framing and choices framing choices the way the question is asked influences the answer we give: Gain vs. losses -​ framing outcome as a gain → leads to risk averse choices -​ framing outcome as loss → risk averse choices Choosing vs rejecting -​ aking which would you choose instead of wich would you reject (between two) also lead to different trains of thought! ( a maioria prefere award e deny custody pra msm pessoa? N tem como ser o melhor e pior ao msm tempo) -​ he presentation or wording of identical choices can lead to different decisions -​ Should the UK remain a member: yes/ no. VS. what should UK do: leave/stay → essa é less biased! quaestio framing Impact on research quality and market research -​ small changes can lead to big differences in responses -​ ex: Do you prefer Coke or Pepsi?” vs. “Which soft drinks do you purchase?” Business implications: -​ Cost of biased data: Product development mistakes, Misaligned marketing strategies, Missed market opportunities Common Question Biased questions: Framing Pitfalls -​ Assumptions: “Why do you prefer our brand?” -​ Leading Questions: “Don't you agree our service is excellent?” -​ Double-Barreled Questions: “Rate our product's quality and price” - n pode ser 2 numa só -​ Loaded Language: “How much time do you waste on social media? - mto negativo, tentar ser neutro Prospect theory accounts for choices under uncertainty: 1)​ Reference points matter! - dependendo de qunato vc ja tem no comeco, uma choice pode paracer boa ou nao. People evaluate outcomes relative to a reference point (gains and losses) rather than in absolute terms! (Point of view matters) 2)​ Loss Aversion: Losses loom larger than equivalent gains 3)​ Risk Behavior changes for loss and gains : For gains, people are generally risk-averse (concave curve). For losses, people tend to be risk-seeking ( convex curve, esperança é a ultima que morre) 4)​ Diminishing Sensitivity: The impact of changes diminishes as they move further from the reference point (o grafico) 5)​ Probability Weighting: People overestimate the likelihood of rare events and underestimate the likelihood of moderate to high-probability events. challenges and Challenge: opportunities -​ difficult to figure out the reference point in real life  Opportunity -​ frame it in a way ppl will do what you want, ex framing tax relief as a bonus instead of relief so economy is boosted Framing and Improving your decisions improving Identify your reference point Consider alternative reference points decisions Play devil’s advocate Applied marketing cases are we too -​ People experience time loss more painfully than time gains impatient? -​ Same journey framed as "scenic route” vs "slower option" changes perception -​ Transport models assume time in transit s always negative -​ Default assumptions about speed being better can limit consideration of alternatives ex -​ framing tudo de um jeito especifico pra give the thought they wan you to have → tentado frame it de um jeiro que vc n tem mais uncertainty na escolha (ex uber mostrando o desnho do carro tira uncertainy, msm q ainda demore mto pro carro chegar) -​ usar rabatt sempre te faz gastar mais na vdd -​ tall sounds better then small -​ good things come to those who wait -​ colocando o nome de carbon tax como “climate action tax” ⇒ frame your message in specific ways to get to your target audience! digital and social media evolution of framing -​ traditional media = One message many channels, Fixed message, Mass audience, Delayed feedback (surveys, sales), Single pre-test -​ Digital media = Many messages many channels, Dynamic adaptation, Personalization, Instant metrics (clicks, views, shares), Data-driven A/B testing o que fazem hj: 1)​ Personalized framing - agr spotify te da a sua top 100, antes todo mundo tinha a msm no radio 2)​ brands change framing across different platforms to cath that specific type of user no face é “give health pra sua familia” no tik tok é “POV: When your Peloton instructor becomes your daily therapy session” 3)​ Algorithms for framing: -​ highly personalized framing, suggestions, movie thumbnails -​ Data-Driven Frame Selection: Real-time testing of multiple frames + Performance metrics guide message evolution -​ New Framing Dynamics: Engagement metrics create rapid feedback loops + Community interaction shapes frame evolution Framing and Ethics Does it empower user choice? Is the framing transparent? Who benefits - user or platform? Are vulnerabilities being exploited? 8: choices and nudges - choice architecture Choice architecture definition = careful design of the environments in which people make choices  -​ this includes Number of options, How they are presented and The presence of defaults -​ there is no neutral design! -​ anyone presenting choices is a choice architect, such as store designers, website developers, and HR managers Nudging = subtle changes in the choice architecture that alter people's behavior predictably without forbidding options or changing economic incentives -​ They are: Easy to avoid, Transparent, Preserve choice freedom, Should benefit the decision maker -​ They are NOT: Rational persuasion, Mandates or bans, Economic incentives (taxes, subsidies), Direct information provision, Designing better -​ pessoas estão em cognitive overload (limited attention, choice overload, decision fatigue, biases) and decisions dont always make the best decisions -​ small details can have big effects -​ choice architecture should improve decision making: -​ make easier decisions -​ reduce errors -​ improve outcomes -​ preserve freedom of choice number of options -​ com mtas options, a gtn entra em cognitive overload. There is a need to balance the preference patch with greeted cognitive load -​ Balance (nr ideal de options) depende do customer: se tem willingness to engage in the choice process, qual a nature da decision and contect (se é high risk), e individual characteristics (se vc é risk averse ou tem high NFC) (research says between 5-7) current debates tech and decision -​ AI can personalize experiences but also raises challenges of transparency and accountability aids -​ tech can used to help decisions through: -​ help design nudges, search engines, product recommendations, interactive comparisons, automatic personalization, and personalized decision agents -​ pode help com decision inertia e when in doubt and need more info, mas tb tem um trade off between uniqueness X efficiency se vc delega todas suas decisões) -​ personalization vs. pushing into something you dont wnat -​ Digital nudging and privacy concerns, especially with apps and platforms collecting behavioral data - consent strategies? -​ Individual vs. corporate vs. public interest: Private companies optimize for profit while governments aim for public welfare - mas o unico beneficiado por choice architecture deveria ser o consumer! -​ Cultural differences in acceptance of nudging - values around individual vs. collective decision-making what can you architecture? the order of options -​ ppl are influenced by the order in which options are presented -​ Primacy Effect: First items receive disproportionate attention and weight -​ Recency Effect: Last items have stronger influence on final decisions -​ ex: ordem de results de uma search engine drastically change click rates -​ ex: Menu design impacts restaurant orders, Product placement in stores/websites influences sales, Survey responses vary based on question sequence default options -​ defaults can be very powerful bc od status qui bias (we like to stick to the current situation -​ ex: organ donations, gym memberships that auto renew, software settings and newsletter subscriptions we never change or cookies! task structure -​ Single vs. serial choices (layout de escolhas -​ ex laptop with fixed specifications vs. customizing a laptop - ate o nr de choices em cada opção possíveis) muda nossa escolha -​ Configurators (ex building your own car) -​ giving a starting solution can help people a lot when customizing, ir attribute by attribute do zero é mto cansativo e ppl end up making less e gostando less das customizações, vai bem mais rapido -​ Filters -​ user-initiated “ppl q bought isso alo got..” , collaborative, progressive “dating apps e job engines que add filters”  -​ Sorting mechanisms -​ Stopping rules (when do I stop searching for better options?) -​ external rules (time limit pra comprar, minimum price pra shipping) vs decision aids -​ Formatting Grouping and Grouping partitioning -​ o jeito q as coisas estao agrupadas influences choices -​ linked to a bias in decision making: Mental accounting = categorize different types of expenditures mentally, e tentam balancear os gastos entre essas categorias Choice partitioning -​ ppl try to equally distribute things - tentam pegar um de cada categoria ⇒ o jeito que products (choices) sao labelled e grouped influencia o que vai ser escolhido (ex qual e quantos de cada categoria!, nao apenas como opcoes soltas) é meio que o supermarket path que te influência, o jeito q as coisas sao posicionadas e grouped Evaluability and People choose between alternatives by weighing pros and cons on attributes → choice architecture could facilitate Comparability that by increasing ease of evaluability and comparability (reducing cognitive effort): -​ giving consumer relevant metrics (nutri-score) (calories per serving, nao no pacote) -​ measurable metrics -​ labels (high/low) -​ Stating explicit consequences (e.g., translating energy consumption into greenhouse gas emissions) o jeito que é mostrado nos ajuda a evaluate essas option and how to evaluate the trade offs we are making Choices Over Time How individuals make choices about the future: a)​ Myopic Procrastination: focar só em comfort agora and procrastinate b)​ uncertainty about the future: ppl struggle to predict future needs or circumstances, affecting their current choices c)​ Hyperbolic discounting: cognitive bias where people disproportionately prefer smaller, immediate rewards over larger, delayed ones. d)​ Planning fallacy: often leads to overly optimistic plans and timelines pq nao planejamos direito e)​ long search process: ppl might allocate all their mental capacity to early, and then have no resources left for subsequent decisions leva á → “depletion of resources”: aceita defaults mais facilmente quando escolha vem dps de uma escolha com mta variety (aka vc ja teve q pensar mto) Como choice architecture tools podem ajudar: a)​ shifting attention to the future - ex com progress bars b)​ Encouraging satisficing by highlighting “second best” options c)​ Limited time offers to overcome procrastination d)​ Break-down into smaller steps e)​ Gamification (like stairs piano) ethics and -​ Nudging should follow Libertarian paternalism: institutions guide while preserving individual choice nudging -​ Tension between individual autonomy and collective welfare Ethical considerations when nudging: -​ Transparency: Digital dark patterns vs. clearly labeled defaults (ex hiding subscription cancellations) -​ Autonomy: Ability to opt-out or choose differently -​ Beneficence: Expert knowledge vs. personal preferences (who does it benefit? Designer should always prioritize the user, colocando healthy food at eye level at the cafeteria/supermarket) -​ Power and Responsibility: Accountability for outcomes, potential for manipulation (it is never neutral to present choices) netflix case -​ Netflix carefully designs its platform to guide user choices, reducing the overwhelming number of options to a manageable selection. -​ The platform uses defaults, such as autoplay and personalized recommendations, to streamline decision-making and keep users engaged. -​ collaborative filtering and content-based filtering to tailor recommendations based on user behavior and preferences - personalization helps users discover content they are likely to enjoy, increasing satisfaction and retention. -​ the strategic placement of content, such as "Trending Now" or "New Releases," influences what users see first, guiding their viewing choices. -​ Visual elements like thumbnails and trailers are optimized to capture attention and encourage exploration  -​ Netflix uses A/B testing and user feedback to continuously refine its choice architecture, ensuring it meets user needs and business goals - iterative process allows Netflix to balance user happiness with operational efficiency. -​ The text discusses the ethical implications of nudging users through choice architecture, emphasizing the need for transparency and user autonomy. -​ Netflix aims to enhance user experience while respecting individual choice, avoiding manipulative practices 9: choice, AI and moral decisions Decision makers and AI AI is reshaping 1)​ in static choice architecture you can do: choice architecture -​ default options, information labels, product placement to guide consumers in predictable ways 2)​ AI introduces personalization! -​ create unique content, build custom interfaces, and give personal recommendations 3)​ Even more advanced: completely engineered environments: -​ AI can learn in real time adapting to user behavior and preferences to optimize engagement! -​ involves attention design, where AI systems strategically capture and maintain user attention. -​ engineered environments include features like Infinite scroll, No time display, Auto-play, Friends active now, Notification design, Like counts, Content recommendations -​ cada pausa, cada scroll, cada like TUDO ta sendo captado e serve pra engineer how to capture attention for longer, adapt to habits e minimize distractions (mas ainda unclear como q fazem isso) -​ nao é mais simple nudging, é um whole environment que learns dynamically and adapts to behavior and data the AI ecosystem -​ AI is completely changing how we make choices -​ starting to anticipate needs e becoming autonomous, adapting to the environment and our behaviors -​ neuralink blurring the barrier between minds and machines? comprehensive nature of AI systems - tudo super interconectado - ta changing how we make choices 1)​ Data collection & storage -​ AI systems constantly gather data through various means. Capta nao só behavioral, mas tb sobre nosso environmental (iphone collect data de apps mas tb de onde vc ta, audio, motion sensors, até health sensors) and interactive sources (neural sources might be next?)→ crucial for training 2)​ Data processing -​ Involves pattern recognition, machine learning, neural networks, and predictive modeling → allow for generation of insights and predictions of future behaviors 3)​ Output systems -​ AI gives outputs through recommendations, automation, and generated content. (ex casa se limpar by itslef sem imput da pessoa pq a casa é smart itself) 4)​ Interactive interfaces -​ include voice, visual displays, and potentially neural interfaces, allowing seamless interaction with users consumers and AI como que as pessoas interagem com AI (text) a)​ Data capture experience: Consumer can feel either served or alienated: -​ AI capability: listening → but through intentional sharing ou passive listening and sensing? -​ Sociological context: surveillance society -​ Alexa never stops listening -​ Consumer can feel either served or alienated:  -​ what can companies to do make you feel more served?: comply with data protection, transparency, getting consent, need to show that consumer get tangible benefits from sharing their data aka added value pra vc como consumer get something in exchange e doesnt feel only like exploitation pq isso de fato enables personalized experience e convenience, da opportunity de self improvement (ex se um fit bit guarda seus dados e vc pode ver) - potential loss of privacy and control over that data b)​ social experience: connected vs alienated -​ AI capability: interacting -​ Sociological context: Humanized AI -​ ex: customer service chatbots virtual assistants AI companions -​ can we really create meaningful connections with AI? -​ -​ pode cirar uma quite positive interaction, mas isso pode backfire! O q acontece se tem update e ele muda? what if it fails? c)​ delegation experience: empowered vs. replaced -​ AI capability: producing -​ Sociological context: transhumanism -​ ex: decision support (recommendations), Partial automation (GPS Navigation), Full automation (smart home) -​ -​ cada vez mais delegamos mais e ficamos confortáveis com isso, deixando ele real fazer coisas on its own, full automation d)​ classification experience: understood vs. misunderstood -​ AI capability: predicting -​ Sociological context: unequal worlds -​ Predictive models categorize consumers and provide personalized predictions or recommendations - Incorrect assumptions by AI can limit exposure to diverse options e classify us with groups we dont identify with  -​ -​ Us tem q ter um balance de providing new inspo, dando oportunidade pra vc descobrir outras new coisas e ainda get relevant recommendations e ter uma personal experience. Values and moral choices Embedded values -​ AI is not neutral -​ program can reflect different values, such as time efficiency, cost savings, environmental considerations, or user experience -​ ex: um AI médico pode ter como values certos padrões de Patient comfort, Cost optimization, Treatment aggressiveness, Risk tolerance - o que ele vai escolher? Kohlberg’s Stages of how individuals progress in their moral reasoning: Moral Development 1)​ Pre-conventional Level: mostly kids - morality is shaped by external consequences rather than internalized ethical principles. Stage 1: Obedience and Punishment Orientation: Moral decisions are based on avoiding punishment. Stage 2: Self-Interest Orientation: Right actions are those that serve one's own needs or interests. 2)​ Conventional Level: teens and mosts adults - people follow societal norms and rules to maintain relationships and social order Stage 3: Interpersonal Accord and Conformity: Good behavior is what pleases others and gains approval. Stage 4: Authority and Social Order Maintaining Orientation: Emphasis on obeying laws and respecting authority to maintain social order. 3)​ Post-conventional Level:not everyone gets here, Morality is guided by abstract principles and personal ethics rather than external rules Stage 5: Social Contract Orientation: Laws are seen as social contracts that should be changed when they do not promote general welfare. Stage 6: Universal Ethical Principles: Moral reasoning is based on abstract reasoning using universal ethical principles. Teaching AI morals -​ AI can be programmed to follow ethical guidelines, such as copyright compliance or promoting inclusive language (social conformity) -​ Systems like autopilots maintain legal standards (law and order morality), while GPS systems might offer eco-friendly routes, considering broader ethical impacts (social contacts/ethics) 10: Guest lecture Romolo E-commerce Trends: Understanding current trends in e-commerce, such as personalization, mobile commerce, and the use of AI in customer interactions. Digital Marketing Strategies: Insights into effective digital marketing techniques, including social media marketing, search engine optimization, and content marketing. Consumer Behavior: How consumer behavior is changing in the digital age and strategies to adapt to these changes. Technology Integration: The role of technology in enhancing customer experience and operational efficiency in e-commerce.  ⇒ These insights can help you understand the practical applications of course concepts in real-world business scenarios 11: Negotiations Common Biases 1)​ Mythical Fixed Pie: The assumption that one party's gain is another's loss, which can limit creative in Negotiations solutions. 2)​ Framing of the Negotiator Judgment: How offers and outcomes are presented can influence risk perception and decision-making. 3)​ Escalation of Conflict: The tendency to continue investing in a failing course of action due to sunk costs. 4)​ Overestimating your own Value: Believing one's contributions are more valuable than they are, leading to impasses. 5)​ Self-Serving Biases: Interpreting information in ways that support one's own interests. 6)​ Anchoring: The influence of the first offer on subsequent negotiations. ⇒ The outcome one achieves in a negotiation is rarely inevitable. In fact, in most negotiations, a wide array of outcomes are possible! ⇒ The lesson? Negotiations should be trained, and the decisions and behaviors of each negotiator matter Schools of Thought in Negotiation Theory Game theory -​ assumes rational behavior, mathematical -​ All agents act perfectly rational -​ It is purely prescriptive, providing strategies based on the assumption of perfect rationality and complete knowledge. -​ pro: prescriptive (aka can predict, given bounded rationality) -​ con: All options and associated outcomes for every possible combination for each person -​ must be known. All agents act perfectly rational Prisoner's dilemma: -​ Even though it would be better to stay silent, both parties will confess (Nash equilibrium), as it is the rational choice to minimize potential loss, even though mutual cooperation would lead to a better collective outcome. -​ aka vamos fazer o que é melhor pra nos mesmos ali na hora → The dilemma highlights the tension between short-term gains and long-term benefits, emphasizing the importance of strategic thinking and cooperation in decision-making. Decision analytic -​ considers biases and real-world behaviors, psychological approach -​ bounded rationality, people make mistakes -​ It is both prescriptive and descriptive, offering advice based on real-world behaviors and likely actions of the other party. -​ Pro: more realistic, included physiological component. -​ Con: less prescriptive Raiffa Key = Combining Prescriptive (what they should do) with descriptive (what they actually do) views. Players: 1)​ Focal Negotiator / Prescriptive (what should do) -​ This is the party for whom the negotiation strategy is being developed. The focal negotiator is the one receiving prescriptive advice on how to approach the negotiation. -​ The focus is on optimizing the focal negotiator's outcomes by understanding their own interests, alternatives (like BATNA), and strategies. -​ The goal is to provide the best possible advice to the focal negotiator based on a realistic assessment of the negotiation scenario. -​ ex: -​ Calculate precise valuation -​ Plan your negotiation moves -​ Set clear walkaway points -​ Prepare rational arguments  2)​ Opponent / descriptive / what actually do -​ The opponent is the other party in the negotiation, whose behavior and strategies are analyzed descriptively. -​ Understanding the opponent involves predicting their likely actions, interests, and potential biases. -​ This analysis helps the focal negotiator anticipate challenges and opportunities in the negotiation. -​ ex: -​ They may be emotionally attached to their company -​ They might anchor on an initial / unrealistic price -​ Personal relationships might matter more than numbers -​ Pride might prevent them from accepting certain terms decision-analytic approach to negotiation emphasizes understanding and analyzing three key sets of information: 1)​ Each party’s alternative to a negotiated agreement (BATNA) 2)​ Each party’s set of interests 3)​ The relative importance of each party’s interests 1) BATNA BATNA = Best alternative to a negotiated agreement, aka best course of action you can take if negotiations fail and an agreement cannot be reached. It represents your fallback option. → Leads you to your minimal acceptable outcome Why is BATNA important: a)​ Knowing your BATNA gives you leverage in negotiations. It helps you understand your minimum acceptable terms and strengthens your negotiating position. b)​ Having a clear BATNA boosts your confidence, as you know you have a viable alternative if negotiations do not go as planned. c)​ Preparation: Understanding your BATNA requires thorough preparation and analysis of your options, which can lead to more informed and strategic negotiation tactics Importance of data (for your BATNA): a)​ Precision Over Assumptions: Data allows negotiators to quantify values precisely rather than relying on assumptions. This precision helps in making informed decisions and setting clear objectives. b)​ Hidden Value Discovery: Analyzing data can uncover hidden opportunities and values that might not be immediately apparent. This can lead to more creative and beneficial negotiation outcomes. c)​ Predictive Power: Data can be used to predict trends and outcomes, helping negotiators anticipate the other party's moves and prepare accordingly. d)​ Clear Walk-Away Points: With precise data, negotiators can establish clear walk-away points, knowing exactly when a deal is no longer beneficial compared to their BATNA. 2) the interest of Often there is a difference between parties' stated positions ≠ and their Interests! Position ≠ interest! each party (not position!) Position = explicit demands Interest = underlying reasons, needs, or motivations behind the stated positions. They represent what the party truly values or seeks to achieve. Understanding the difference is crucial because: a)​ understanding what each party truly values allows for exploring potential trade-offs and creating value in negotiations b)​ Interests allow for more flexibility in negotiations, as there may be multiple ways to satisfy them beyond the initial positions. 3) The Relative Knowing how important each interest is to the parties involved helps in structuring trades and concessions Importance of Each effectively. Parties’ Interests -​ Understanding the relative importance allows negotiators to structure trades that maximize value for both parties. For example, one party might concede on a less important issue in exchange for a more critical one. -​ Recognizing the relative importance of interests helps prevent deadlocks by ensuring that negotiations  focus on what truly matters to each party. Bargaining zone a)​ situations with positive bargaining zone → there is a solution! the worst situation is when there is a solution and bc of greed a settlement does not happen 3 key skills: -​ determine your BATNA -​ determine other party’s BATNA -​ aim for a resolution that is barely acceptable to the other party b)​ situations with negative (no) bargaining zone → don't lead to settlement! -​ determine your BATNA -​ learn to say No Claiming and ideal negotiation strategy: creating value -​ your objective is to both claim value and create value! (most ppl struggle to do both) -​ claim: get a part for yourself -​ create: expanding the available value or resources so that all parties can benefit -​ Effective negotiators often aim to create value first by identifying shared interests and opportunities for mutual gain. Once value is created, they then focus on claiming their fair share. By definition, one-issue negotiations (i.e., only salary) are Value Claiming negotiations only… ⇒ Through the process of identifying and adding issues, parties have the potential to create value! Take Diffeences as opportunities for Value Creation! Five prerequisites and steps: 1)​ mentality: Take differences as opportunities, NOT as a problem 2)​ identify new issues 3)​ Add issues to the negotiation to transform a one-issue negotiation to a multi-issue negotiation 4)​ Focus on issues that parties weigh differently 5)​ Trade issues of differential value: Make concessions at issues being of low value for you but of high value for the other party, and get in return concessions at issues being of high values of you but of lower value for the other party Other aspects of negotiation The power of AI can help with having data to trust on! It can help with: -​ Value prediction: analyzes data in real time, predict future values -​ Hidden value factors: future development potential, trends, projects etc -​ Risk assessment: condition analysis, metrics etc Contingecy ⇒ making the price of an offer contingent on performance! contracts -​ ex: performance bonuses, sucess-based components, sales-based royalties and bets How they help: 1)​ Bets build on differences to create joint value - creating contigent contract based on differing predictions 2)​ Bets help to manage bias - Contingent contracts allow agreements to be formed despite these biases 3)​ Bets diagnose disingenuous parties - rejection of bet reveals bluff pq dai é obv que o preditcion nao era sincere 4)​ bets establish incentives for performance - increasing the parties’ incentive to perform above contractually specified levels” strategies to collect 1)​ build trust and share information (may sound coterintuitive as betray aversion is big concern) info on othe party 2)​ Ask questions - bc u have to understand the interest of the other party as well as possible ex de good questions: How are you going to use our products? What would an ideal supplier do to make its product attractive to you? How can we make our offer better than that of our competitor? 3)​ Strategically disclose info - share incrementally, but never your reservation price! 4)​ negotiate multiple issues simultaneously 5)​ make multiple offers simultaneously - they should all be equally attractive to you. Avoid unfavourable anchors, Shows the other party that you are “customer” oriented and strive for reaching a solution. Helps you to receive implicit and valuable information from the other party. If offers declined: Ask which package was most attractive and how this can be reworked to make it more attractive 6)​ search for post-settlement settlements - “Freeze”/”Save” the agreement and search without any hassle if there is a better solution (Pareto-superior agreement) Key Takeways ​ Understanding Interests vs. Positions: Focus on underlying interests rather than just stated positions to find  mutually beneficial solutions. ​ Creating and Claiming Value: Aim to create value by exploring shared interests and then claim your fair share through strategic negotiation. ​ BATNA: Know your Best Alternative to a Negotiated Agreement to strengthen your position and make informed decisions. ​ Data Utilization:Use data to enhance precision, discover hidden values, and set clear negotiation boundaries. ​ Contingency Contracts:Consider using contingency contracts to manage risk and align incentives based on future uncertainties. ​ Framing Effects: Be aware of how framing can influence perceptions and decisions, and use it to your advantage. ​ Preparation and Strategy:Thorough preparation and understanding of both your and the other party's interests are crucial for successful negotiations. 12: Consumer behavior research - Guest lecture Key takeaways 1)​ capture behavior instead of demographics (diz mto mais!) 2)​ start with "why” - crate smart research questions out of the business questions - ask the client why they want to know that so we can get to actual results they want 3)​ check for biases at every stage of research e nao push ppl to answer what you want 4)​ design for action, pense em smart questions que vao derive action, e nao so usar uma methodology so pq é uma fancy methodology right now, ensure insights drive decisions 5)​ consider the context - environment, culture, and temporal factors podem afetar sua research! vc tem q estar aware of it entnder sobre system 1 and 2 te ajuda a fazer pesquisas emlhores — se vc sabe que 95% é system 1, entao vc fica melhor se vc faz questionnaires pensanod que as pessoas nao usam mto rational side vc tb quer ter empirical evidence de vdd! pra entender o o consumer de vdd! tem uma difernca entre o que as pessoas claim to do e o que as pessoas realmente do! vc quer saber o q eles de fato fazem neh - keep epirical observations in the back of your mind eerytime you do reseacrch! 1900 e pouco - faziam coisas “na mao” 70 focus groups aparecem e scanner technology 90 so que tem computador pra fazer online 2000 e 2010 social media aprece e a gnt tem mta mais info! Agr um super automated process pra fazer pesquisas 2020 AI analytics agr ⇒ na vdd o que teve inovação foram os devices! e não necessariamente a técnica - o ponto que ela quer passar é que nao é pra ter medo desses nomes novos tipo “digital forensics” na vdd a base boa sao coisas antigas mesmo , o novo sao os devices synthetic customers - vc “cria” consumers em LLMs e usa eles pra responderem suas consumer research questions em vez de ter que ir perguntar pessoas de vdd vc pode pedir pra LLM responder como um consumer e promt pra ele sera q tem mais eficiência? - ajuda a identificar que perguntas podem ser meio bocós e que outras sao real boas e ainda blurry onde perguntar pessoas de vdd iria ajudar mto! Aka ajuda a refine sua research antes de go into the field , podem ajudar a interview vulnerable pessoas e get in depth answers de quem poderia ser dificil pq podem se explicar melhor (se o LLM for modelled como um consumer assim) however, ajuda com first impressions, mas ainda assim não é o real deal neh, ele pode esquecer os initial  perguntas, faltar um pouco de depth e nuance talvez , nao pega new trends e cultural shifts. Se ele for treinando com um western european industrial demographic ele so vai responder assim, e nao vai te dar insights to other ppl what questions should Paradise Road really be asking about their advertising?: Brand association does this translate into behavior whats the nr 1 objective brand awareness budget allocation how to reach them? Which channel how to stand out form competition what benefits do they want (n so wether they like it) message recall 13: Collaboration and group decisions Group decision making Definition and scope Group decision-making involves collective deliberation or aggregation of preferences to make judgments or choices. Decisions range from simple aggregation (e.g., voting) to fully interactive, consensus-based processes. Advantages ​ Groups often outperform individuals in problem-solving, negotiation, and decision accuracy due to collective knowledge. ​ Majority/plurality-based decisions and aggregation techniques like averaging tend to produce better outcomes in many situations. ​ Diverse perspectives and shared conceptual systems improve decision quality. Limitations ​ Group decisions are not always optimal and can lead to poor outcomes due to biases like social sharedness and the common knowledge effect. ​ Factors like group conformity, unshared information, and egocentric biases can negatively influence decision-making. ​ Group dynamics often amplify initial majority views, which may exacerbate biases if those views are incorrect. common knowledge ​ Groups naturally tend to focus on shared or commonly known information rather than unique information effect held by individuals (the "common knowledge effect") ​ groups that emphasize accuracy or solving the problem (rather than simply reaching consensus) are more likely to consider unique information. Combat with: Delphi Method: ​ Members provide input independently, which is then aggregated and shared without revealing who said what. This minimizes social pressure and ensures unique information is considered. ​ Iterative feedback loops allow members to revise their contributions based on the group's aggregated input​. Nominal Group Technique: ​ Members brainstorm ideas individually before discussing them as a group to prevent early dominance by certain opinions​. Pre-Discussion Information Sharing: ​ Have members present all relevant information before any preferences or decisions are expressed, ensuring all perspectives are considered before forming judgments​- Delay preference sharing until all relevant information has been exchanged - When group members state their preferences early, discussions often focus on justifying those preferences rather than exploring all available information. Transactive memory system ​ Assign specific roles or areas of expertise to group members to ensure that all relevant information is covered. ​ How It Helps: By clearly defining who is responsible for certain knowledge, groups can systematically incorporate unique insights during discussions​. Tech and Memory aids ​ Group members may forget or overlook key details during discussions.Maintain a visible record of all  information discussed, ensuring unique insights are not lost​ ​ A neutral party can help guide discussions to ensure that all members have an opportunity to share their view Processes and ​ Aggregation methods (averaging, medians) and models like Social Decision Schemes (SDS) describe techniques group preference reconciliation. ​ Techniques like the Delphi method enable structured, anonymous input sharing to reduce biases and improve outcomes. ​ Prediction markets and Judge–Advisor Systems (JAS) are other methods for combining insights while managing social influence. Factors for effective ​ Encouraging information exchange and reducing conformity pressures. group decisions ​ Diverse group composition with members who are open to differing viewpoints. ​ Shared goals, high epistemic motivation, and pro-social orientation among group members. ​ Groups work best when motivated by accuracy, not self-interest, and when shared biases are minimized. ​ Groups will also be wiser when they are composed of wiser members ​ allowing members to exchange information and ideas tends to do little harm and can allow groups to take advantage of particularly good ideas uniquely held by few members ​ When group members all share a biased representation of the decision environment, group discussion tends to exacerbate such biases ​ Groups also can be unwise when they make decisions that have direct implications for the well-being of the group - The only incentive should be to be as accurate as possible ​ Finally, groups are wiser when their members exchange all of the available information rather than just focusing on information they all share - Key differences Aggregation: aggregating vs group discussion ​ Definition: Limited or no interaction among group members.Individual preferences or judgments are collected and combined mathematically (e.g., voting, averaging). Members may not even be aware of others’ inputs or decisions. ​ Info sharing: Minimal opportunity for information exchange. Relies on members' individual knowledge and perspectives without collective deliberation. Unshared or unique information held by individual members is often lost. ​ Bias and social influence: Reduces the risk of conformity pressures or dominance by high-status or vocal individuals. Decisions are often based solely on numerical or statistical methods (e.g., majority/plurality rules, averaging), which can limit the influence of any one member. ​ Efficiency: Useful for simple, large-scale decisions like voting in elections or prediction markets.Works best when members have diverse but unbiased judgments. Group discussions: ​ Definition: High levels of interaction, with active communication, debate, and information exchange among members. Members collaboratively work toward a shared understanding or consensus. ​ Info sharing: Encourages sharing of diverse knowledge and perspectives. Allows for clarification, integration, and collective analysis of information. However, shared or "common" information tends to dominate discussions, leading to potential biases (info q so uma pessoa tem nao vao ser principal) ​ Bias and social influence: Can be influenced by social dynamics, such as conformity pressures, groupthink, or dominance by vocal or high-status individuals. However, it also allows for minority opinions to potentially influence the majority if the arguments are strong and supported by evidence. ​ Efficiency: Best suited for complex or ambiguous problems that require deeper analysis.Vulnerable to shared biases or poorly managed group dynamics, which can reduce decision quality. Minorities ​ can change group decisions when their position is demonstrably correct and If it resonates with shared values, beliefs, or assumptions held by the group aka when they have shared task representations (common cognitive frameworks or values) and shared motivations - also when the right answer is unclear ​ "Truth Wins" and "Truth Supported Wins" models suggest that minorities holding the correct solution can influence groups when their reasoning aligns with a shared conceptual framework or knowledge base ​ Minority members often possess unique information or insights not shared by the majority ​ Diverse groups are more open to considering minority viewpoints. Inner crowd Inner crowd -​ Most helpful with numeric values -​ comes from the “wisdom of crowds” concept -​ ppl together always get closer to the actual value, unless it is not an extreme (like 95%) -​ averaging one person’s guesses will come closer to the actual value -​ multiple guesses from a single person do suffer from a major drawback. They are typically quite similar to one another, as people tend to anchor on their rst guess when generating a second guess. -​ Methods that try to improve: Sleep on it + dialectical bootstratping (ask to base second guess on different assumptions) -​ taking the perspective of a friend you disagree with could help your own estimates look more like the aggregate estimates of a diverse, independent group of people. -​ taking the perspective of a friend you disagree with improves your guess! -​ being prompted to incorporate a disagreeing perspective can lead to more diverse and independent estimates, seeding the conditions necessary to make the wisdom of the crowds effective within an individual. 14: collaboration and augmentation & decision strategies Group decision making social nature of -​ most decisions have social component decision -​ decisions are deeply routed in social interaction group decision -​ better problem solving advantage -​ more accurate forecasts -​ improved negotiation outcomes -​ enhanced creativity -​ negative side: biases! Key dimensions of Dimensions to characterize group decision making processes: decision making: 1)​ Shared motivations! -​ Members develop a collective mindset that prioritizes group welfare -​ Can lead to positive outcomes and decision biases -​ Ingroup bias leads to protective and enhancement-oriented behaviors -​ There are two types of shared motivation and both matter: i)​ Epistemic motivation = desire to reach the most accurate outcome, drive for accuracy and thorough analysis - pushes people to put more info on the table, haver longer discussion, discover new info that hasn't been shared/discussed yet ii)​ pro-social motivation = desire to work together, share info and help group succeed Vs. pro-self orientation → can backfire and lead to over protective behavior (not to harm others, but to protect group) iii)​ Optimal performance = High epistemic + pro-social motivation 2)​ Interaction Dynamics: how we interact, talk, vote affects the group dynamic a)​ Communication patterns (written or verbal) b)​ Interaction intensity (no interaction to intense group discussion) c)​ interaction type (is it real time, asynchronous, through shared documents, etc) 3)​ Decision control: how is the final decision reached? how do we agree, share, vote, evaluate the options a)​ control mechanisms - specific tools or processes: performance reviews, audits, and feedback systems b)​ control strategies - broader plans or approaches used to guide decision-making and behavior towards achieving specific goals: framework or methodology for making decisions, such as using decision-analysis tools, setting clear objectives, and prioritizing actions based on strategic importance Interaction