Predictive modelling - PDF Free Download

Predictive modelling Stanley Greenstein Stanley.Greenstein@juridicum.su.se Predictive Modelling Länk: http://su.divaportal.org/smash/record.jsf?pid=diva2% 3A1088890&dswid=-1010 2017-11-28 Stanley Greenstein, LL.D. 2

Plan for this presentation: 1. Predictive modelling (technology) 2. Application 3. Harms 4. The law and empowerment 2017-11-28 Stanley Greenstein, LL.D. 3 1. Predictive modelling 2017-11-28 Stanley Greenstein, LL.D. 4

Predictive modelling Mathematics, statistics, machine learning, AI Algorithms identify patterns in data that are invisible to human beings (correlations = knowledge ) This knowledge is applied in models that control/determine our contact with the digital environment Models can identify and predict human behaviour That not why! Increasingly being used in digital decision systems 2017-11-28 Stanley Greenstein, LL.D. 5 Steven Finlay A predictive model captures the relationships between predictor data and behaviour, and is the output from the predictive analytics process. Once a model has been created, it can be used to make new predictions about people (or other entities) whose behaviour is unknown. 2017-11-28 Stanley Greenstein, LL.D. 6

Witten & Frank A scientist s job is to make sense of data, to discover the patterns that govern how the physical world works and encapsulate them in theories that can be used for predicting what will happen in new situations. The entrepreneur s job is to identify opportunities, that is patterns in behaviour that can be turned into a profitable business, and exploit them. 2017-11-28 Stanley Greenstein, LL.D. 7 Netflix 2017-11-28 Stanley Greenstein, LL.D. 8

Netflix 2017-11-28 Stanley Greenstein, LL.D. 9 Decision tree 2017-11-28 Stanley Greenstein, LL.D. 10

Neural network 2017-11-28 Stanley Greenstein, LL.D. 11 2+2=3,9...and that s OK 2017-11-28 Stanley Greenstein, LL.D. 12

Manhole covers i NY Problem of exploding manhole covers in NY Con Edison given the task of solving the problem Inspections based on chance 152 000 km wiring & 51 000 manhole covers Interested in that and not why Use the data Top 10 % on list - 44 % of manholes in really bad shape 106 factors e.g. age of wiring, previous problems Correlation and weight of these correlations Mayar-Schönberger, Victor & Cukier, Kenneth, Big Data, 2013 2017-11-28 Stanley Greenstein, LL.D. 13 Correlation 3 1 2 1 2 2017-11-28 Stanley Greenstein, LL.D. 14

Flight Risk 2017-11-28 Stanley Greenstein, LL.D. 15 BIG DATA 2017-11-28 Stanley Greenstein, LL.D. 16

Big Data cont. Prosthesis Conway, Andrew and Eckersley, Peter, When Does Law Enforcement s Demand to Read Your Data Become a Demand to Read Your Mind?, Communications of the ACM, 2017. 2017-11-28 Stanley Greenstein, LL.D. 17 2. Applications 2017-11-28 Stanley Greenstein, LL.D. 18

Spotify 2017-11-28 Stanley Greenstein, LL.D. 19 Instagram 2017-11-28 Stanley Greenstein, LL.D. 20

Dating 2017-11-28 Stanley Greenstein, LL.D. 21 Facebook 2017-11-28 Stanley Greenstein, LL.D. 22

Health and medicine 2017-11-28 Stanley Greenstein, LL.D. 23 Facebook 2017-11-28 Stanley Greenstein, LL.D. 24

Facebook 2017-11-28 Stanley Greenstein, LL.D. 25 Sanebox 2017-11-28 Stanley Greenstein, LL.D. 26

China 2017-11-28 Stanley Greenstein, LL.D. 27 Facial recognition 2017-11-28 Stanley Greenstein, LL.D. 28

Facial recognition the upper lip curvature is on average 23.4% larger for criminals than for noncriminals. the distance d between two eye inner corners for criminals is slightly shorter (5.6%) than for non-criminals 2017-11-28 Stanley Greenstein, LL.D. 29 Facial recognition 2017-11-28 Stanley Greenstein, LL.D. 30

Facial recognition 2017-11-28 Stanley Greenstein, LL.D. 31 Law enforcement 2017-11-28 Stanley Greenstein, LL.D. 32

3. Harms 2017-11-28 Stanley Greenstein, LL.D. 33 Legal harms Privacy/Data-protection (integritet/dataskydd) Reputation (rykte) Autonomy (självbestämmande, autonomi) Discrimination (diskriminering) Manipulation (manipulation) De-individualization* Stereotyping* (stereotyper) Stigmatisation* (stigmatisering) Self-censorship* (självcensur) 2017-11-28 Stanley Greenstein, LL.D. 34

The filter bubble 2017-11-28 Stanley Greenstein, LL.D. 35 The Daily Me 2017-11-28 Stanley Greenstein, LL.D. 36

Influencing human behaviour? 2017-11-28 Stanley Greenstein, LL.D. 37 How to influence human behaviour? 45% of daily human activity is carried out via habit Humans think using 2 systems: System 1: intuitive, automatic (rapid, feels instinctive, feels like we are not thinking) System 2: reflective, rational (deliberate, self-conscious) Certain decisions system 1 works but is tainted with a certain bias Short-sighted, optimistic, mental shortcuts, probability Manipulation is usually directed against system 1 (or overload system 2) E.g. Subliminal advertising 2017-11-28 Stanley Greenstein, LL.D. 38

Influence human behaviour https://www.aftonbladet.se/nyheter/a/orp51/har-ar-golvet-som-far-manniskor-att-sakta-ner-i-korridoren 2017-11-28 Stanley Greenstein, LL.D. 39 2017-11-28 Stanley Greenstein, LL.D. 40

2017-11-28 Stanley Greenstein, LL.D. 41 2017-11-28 Stanley Greenstein, LL.D. 42

2017-11-28 Stanley Greenstein, LL.D. 43 2017-11-28 Stanley Greenstein, LL.D. 44

4. The traditional law & empowerment 2017-11-28 Stanley Greenstein, LL.D. 45 What does the law say? ECHR (Art 8: Right to private life) Data Protection 2017-11-28 Stanley Greenstein, LL.D. 46

ECHR In the opinion of the Commission, however, the right to respect for private life does not end there. It comprises also, to a certain degree, the right to establish and to develop relationships with other human beings, especially in the emotional field for the development and fulfilment of one s own personality. X v. Iceland (Application no. 6825/74), Commission Decision, 18 May 1976. 2017-11-28 Stanley Greenstein, LL.D. 47 ECHR Article 8 also protects a right to personal development, and the right to establish and develop relationships with other human beings and the outside world Although no previous case has established as such any right to self-determination as being contained in Article 8 of the Convention, the Court considers that the notion of personal autonomy is an important principle underlying the interpretation of its guarantees. Pretty v. United Kingdom (2002) 35 EHRR 1, para. 61. 2017-11-28 Stanley Greenstein, LL.D. 48

ECHR The Court reiterates that the concept of private life extends to aspects relating to personal identity Furthermore, private life, in the Court s view, includes a person s physical and psychological integrity; the guarantee provided by Article 8 of the Convention is primarily intended to ensure the development, without outside interference, of the personality of each individual in his relations with other human beings There is therefore a zone of interaction of a person with others, even in a public context, which may fall within the scope of private life. Von Hannover v. Germany, para. 50. 2017-11-28 Stanley Greenstein, LL.D. 49 General Data Protection Regulation 2016/679 GDPR 2017-11-28 Stanley Greenstein, LL.D. 50

GDPR... nobel idea, but... Complex and vague Same data protection principles as the Directive (DPD) Art. 22, Recital 63 GDPR Consent Effective in relation to modern day technology? Does not relate to correlations 2017-11-28 Stanley Greenstein, LL.D. 51 Recital 63, GDPR Every data subject the logic involved in any automatic personal data processing and, at least when based on profiling, the consequences of such processing That right should not adversely affect the rights or freedoms of others, including trade secrets or intellectual property and in particular the copyright protecting the software. However, the result of those considerations should not be a refusal to provide all information to the data subject. Where the controller processes a large quantity of information concerning the data subject, the controller should be able to request that, before the information is delivered, the data subject specify the information or processing activities to which the request relates. 2017-11-28 Stanley Greenstein, LL.D. 52

Correlations 3 1 2 1 2 2017-11-28 Stanley Greenstein, LL.D. 55 EMPOWERMENT 2017-11-28 Stanley Greenstein, LL.D. 56

Is the traditional law empowering enough? 2017-11-28 Stanley Greenstein, LL.D. 57 Empowerment 2017-11-28 Stanley Greenstein, LL.D. 58

Soft law 2017-11-28 Stanley Greenstein, LL.D. 59 Empowerment 2017-11-28 Stanley Greenstein, LL.D. 60

Chatbot 2017-11-28 Stanley Greenstein, LL.D. 61 Digitocracy? A new set of more complex governance mechanisms assuring public accountability for online power held by state and nonstate actors through the creation of new checks and balances among a more diverse group of players than democracy s traditional grouping of a representative legislature, executive branch, and judiciary. Reidenberg, Joel R., Law and Technology Digitocracy, Communications of the ACM, Sept., 2017. 2017-11-28 Stanley Greenstein, LL.D. 62

Digitocracy cont. The mechanisms for states, private actors and citizens to co-exist as rule-makers in the networked society are likely to be defined in unexpected ways incorporating notions of federalism, multistakeholder governance, and subsidiarity. These tools will draw the boundaries of rule-making authority among the state actors, platform operators, corporate organizations and empowered users. Each actor, whether state or non-state, has an important role to prevent the overreaching of the other actors. In essence, Digitocracy constructs a more multifaceted set of interwoven checks and balances to establish limits on the powers of both state and non-state actors and a reliance on both to protect the public good. Reidenberg, Joel R., Law and Technology Digitocracy, Communications of the ACM, Sept., 2017. 2017-11-28 Stanley Greenstein, LL.D. 63 2017-11-28 Stanley Greenstein, LL.D. 64

The predictive modelling process 2017-11-28 Stanley Greenstein, LL.D. 65