Introducing 

Prezi AI.

Your new presentation assistant.

Refine, enhance, and tailor your content, source relevant images, and edit visuals quicker than ever before.

Loading…
Transcript

“The key to our product is the algorithms, and they’re proprietary,” said Jeffrey Harmon, Northpointe’s general manager.

“We’ve created them, and we don’t release them because it’s certainly a core piece of our business.

It’s not about looking at the algorithms. It’s about looking at the outcomes.”

intellectual property opposition

How do we reconcile the legitimate intellectual property rights exercised over algorithms with the explainability issues that arise when they are deployed in the realm of governmental or public interest decision-making?

semantic sophistication

Deep learning, reinforcement, neural networks

multiple layers of representation

Pasquale, Diakopoulos

  • audit input data
  • monitor results
  • control processing rules
  • ban certain kinds of sensitive automation processes
  • regulate sensoring

AI as a set of techniques across

different applications

Law is important, but other governance tools

have to be considered

re-coding of the rule of law

responses might vary across sectors and jurisdictions

What does society

have to do with all that?

Further Developments

Specialiste

We all know that a constitutional or EU Law test will not ban automation just because there is no man in the loop, but only over prejudice (even if potential) or concrete harm

No framework for impact assessment, as in Environmental or Competition

de

- intellectual property rights over algorithms and data sets with significant public relevance

l'animation

GDPR, Art. 22

- Environmental Law And Competition Law use impact assessment tools that might help

A

PLPD Brasil, Art. 19

transparency

accountability

Why now?

fairness

complexity

explainability

What?

and not before?

automated law-enforcement

legal

usability

"the extent to which a product can be used by specified users to achieve specified goals with effectiveness, efficiency, and satisfaction in a specified context of use."

ISO 9241-11, Geneva, 1988

Moment of decision (high semantic effort)

x

Moment of enforcement (lower semantic effort)

  • maturity of the technology
  • social impact (depth of potential transformations)
  • legal impact

potential applicability

Identify stages of technological support for law-enforcement procedures

(manual, instrumental, automation)

Teleonomic

X

Teleological

real-time law enforcement (man in the loop)

x

unmanned law-enforcement (man out of the loop)?

Moment of enforcement (lower semantic effort)

Teleonomic

instrumental or automated

Solutions are unstructured, not systematic, developed case by case and technology by technology

Maybe instances are not yet legally significant, but adoption and massification will make Law incorporate them

real-time or

unmanned

intense, fast, deep

and wide disruption

fully implemented

Lorem ipsum

considerable direct impact

legally sensitive

Cras elementum ultrices diam. Maecenas ligula massa, varius a, semper congue, euismod non, mi. Proin porttitor, orci nec nonummy molestie, enim est eleifend mi, non fermentum diam nisl sit amet erat. Duis semper. Duis arcu massa, scelerisque vitae, consequat in, pretium a, enim. Pellentesque congue

fully implemented

average direct impact

not so legally sensitive

legal order

if we weren't just imagining

Imagine

but now

research project

massive direct impact

legally sensitive

context

fully implemented

massive direct impact

legally critical

approach

under development

considerable direct impact

legally critical

algorithms

in use

sensors

unknown direct impact

legally critical

and

data

from running stock markets to the criteria of my VISA appointments, from my bank credit to the offers I get as a consumer, from my risk assessment by health and insurance companies to which neighborhoods to direct police resources, from being a suspect in a crime to public policy and interest decisions

accountability

compliance

GPS

Ambient Light Sensor

Accelerometer

Air Gesture

Magnetometer

Gyroscope

Proximity Sensor

NFC

ICIL 2016

7th International Conference on Information Law and Ethics ΙCIL

Broadening the Horizons of Information Law and Ethics

Α Time for Inclusion

Pretoria, South Africa

February, 2016

automated law-enforcement

legal

usability

"the extent to which a product can be used by specified users to achieve specified goals with effectiveness, efficiency, and satisfaction in a specified context of use."

ISO 9241-11, Geneva, 1988

Moment of decision (high semantic effort)

x

Moment of enforcement (lower semantic effort)

  • maturity of the technology
  • social impact (depth of potential transformations)
  • legal impact

potential applicability

Identify stages of technological support for law-enforcement procedures

(manual, instrumental, automation)

Teleonomic

X

Teleological

real-time law enforcement (man in the loop)

x

unmanned law-enforcement (man out of the loop)?

Moment of enforcement (lower semantic effort)

Teleonomic

instrumental or automated

Solutions are unstructured, not systematic, developed case by case and technology by technology

Maybe instances are not yet legally significant, but adoption and massification will make Law incorporate them

real-time or

unmanned

How should legal frameworks incorporate the use of sensors, data and algorithmic processes to automate measures of law-enforcement?

iGorithm

You are what you browse

How, if at all, do developers think of legal implications of technology they are developing (empirical gauging)?

Is there a human irreducible semantic core in the act of deciding (does it matter if it is a legal or a non legal decision)?

Which public interests prevail for State-favouring decisions and analysis?

with assess does it imply test? YES

do we take that the technology (robocop style) is always right? NO

would there be any safety clause ( like yelling help me the robot is killing me) CORE OF THE MODEL

is speed always a good thing? NO. VERY GOOD POINT

can automated law enforcement be for any kind of violations> DEFINITELY NOT. PART OF THE JOB IS TRYING TO IDENTIFY WHERE IT FITS AND WHERE IT DOES NOT

will have this legislation have a certain cost? IT WILL. BUT A FEATURE OF AUTOMATION IS THAT MARGINAL COST IS LOW

the difference between writing a law and a code ( precision) HISTORICALLY HUGE, BUT CURRENT STAGES OF IA AND DEEP LEARNING MAKES IT DEBATABLE

will the law also account for the rare cases that the system has made an error? DL AND IA AGAIN

georreferenced and content-based

  • What rights are at stake?
  • What is their nature? Fundamental ones?
  • Is it necessary to redesign rights themselves?
  • Are the ones which are widely acknowledged enough?
  • Is it necessary to create new ones (RTBF, delist)?

Prezi

Why Law?

Where?

Cláudio Lucena

it is by all means unacceptable to despise that the development and the incorporation of digital technologies in everyday human life both have paramount impact over fundamental and personality rights, as well as over other non-economical values duly protected by Law."

Cláudio.Lucena@uepb.edu.br

Descriptive

Concentration on structural informational elements (sensors, data, algorithms), rather on systems (devices or systems, drones, robots, etc.)

Review the context of the research from a Law and Technology approach, present concepts, examine classifications and interdisciplinary features, establish relationships and explore alternatives, using existing theoretical references to justify the research choices made from within available options

Empirical

Case study to identify characteristics of selected technologies which are currently in use to automate law-enforcement measures, which legal elements are considered and the policy-making process

Traditional legal interests

Holistic view, general theory (as opposed to specific liability, military use, etc.)

Experiment?

regulatory law

consumer protection

property

Product liability

data protection

contracts

Normative

Being possible to identify common features among studied instances, consider the viability of a legal dimension of the concept of usability in the realm of automated law-enforcement, as an element of a framework to assess the legality, maturity and adequacy of technological solutions deployed into legal coercive measures

intellectual property

New legal interests

intergenerational data protection framework

behavioral nudging of the networked self

deep learning normativity

reciprocal transparency

usability

delisting

automated law-enforcement

internet governance

global constitutionalism

world scale data operators

algorithmic accountability

international jurisdiction

non deterministic

but

nature

legally relevant to govern algorithms

No

challenges

measures

Fine-tuning

Jacky Alciné

Word Embedding

YouTube ContentID

How is it possible to design an accountability framework when deploying deep learning mechanisms? (auto-feedback, decision, even if not necessarily legal)

Jurisdictional compatibility and

interoperability of automated coercion

Private enforcement (delist, RTBF)

X

Public enforcement

(administrative instances X court jurisdiction)

Compulsory mass sensoring legal?

Shift from compliance to individual monitoring?

What interests can it benefit?

Under which conditions?

Which jurisdictions would adapt?

(comparative approach)

Which scopes of law would fit in?

(case approach, law and economics)

"(...) the legal interest in digital technologies originates out of the simple and obvious insight that their effects and the behavioral transformations they induce in the social tissue can – and already do – interfere with the sphere of rights and obligations of both legal and natural persons.

That alone is enough for Law and its structures to care."

Collective Rights and Digital Content, Cláudio Lucena, Springer, 2015.

VOTRE TIRE ICI

Cras elementum ultrices diam. Maecenas ligula massa, varius a, semper congue, euismod non, mi. Proin porttitor, orci nec nonummy molestie, enim est eleifend mi, non fermentum diam nisl sit amet erat. Duis semper. Duis arcu massa, scelerisque vitae, consequat in, pretium a, enim.

Créer votre lien : http://www.prezcreation.com

WS #264 Automated Guardians of the Good? Algorithms impact in the exercise of rights

Catherine García, Cláudio Lucena, Martín Silva, Rachel Pollack, Sizwe Snail

Geneva, Switzerland

December 20th, 2017

par http://www.prezcreation.com

3D TEMPLATE

growingly harder

ethical issues

not to mention

Learn more about creating dynamic, engaging presentations with Prezi