hyper-LINC #09 : design de l'attention, pouvoir d'agir et neutralité des algorithmes

• Beyond Explainability: A Practical Guide to Managing Risk in Machine Learning Models

FPF and Immuta released the first-ever framework for practitioners to manage risk in artificial intelligence and machine learning models. The joint whitepaper, Beyond Explainability: A Practical Guide to Managing Risk in Machine Learning Models, provides business executives, data scientists, and compliance professionals with a strategic guide for governing the legal, privacy, and ethical risks associated with this technology.
Future of Privacy Forum | 26 juin 2018

Données personnelles et pouvoir d'agir des individus

• Why personal agency matters more than personal data​

Lately a lot of thought, work and advocacy has been going into looking at personal data as a fungible commodity: one that can be made scarce and bought, sold, traded and so on. Good though this might be, it also steers attention away from a far more important issue it would be best to solve first: personal agency. (...) The fact that adtech plants tracking beacons on our naked digital selves and follows us like marked animals across the digital frontier may be a norm for now, but it is also morally wrong, massively rude and now illegal (at least in spirit) under the GDPR.
Doc Searls sur Medium | 24 juin 2018

RGPD et USA? 

• Silicon Valley execs will meet to discuss privacy​

The U.S. has generally approached privacy rules on a sector-by-sector basis, meaning the health care industry has different privacy standards than the financial industry. Tech companies handle data according to their privacy policies and other agreements, such as the Privacy Shield between the EU and U.S. And the FTC makes sure companies stay true to their promises to consumers. (...) "Just because Europe has taken a comprehensive approach doesn't mean our different approach is deficient," Garfield said. "And just because Europe is early doesn't mean it's best or final. But we should always be thinking about how we evolve to make sure consumers have trust in our products."
Axios | 25 juin 2018

Les manipulations des utilisateurs par le design

• How the internet tricks you out of privacy using 'dark patterns' of design

" If you want to know when social media companies are trying to manipulate you into disclosing information or engaging more, the answer is always," he said"​ (...) Professor Hartzog said consumers need to watch out for the tricks and symbols that denote safety and control — the padlock icon being one of the most common — and ask themselves whether they can be relied on. These tricks of perception of privacy can be "dark patterns". A term coined by British designer Harry Brignull, dark patterns are online design choices that obscure and manipulate a website's true intention or function."
ABC science | 30 avril 2018
 

Algorithmes neutres/pas neutres​

• « Ce que les algorithmes voient est conditionné par l'argent »

« Les algorithmes de deep learning ne viennent pas du cerveau d’un génie dans un garage : ils dépendent énormément de la production industrielle de données. (...) Ces algorithmes sont extrêmement dépendants de ces images, ils n’ont rien en interne pour contrebalancer ce qu’ils apprennent. L’étape d’entraînement est donc très importante, elle détermine leur vision du monde car c’est à partir de ces images-là qu’ils vont transposer ce qu’ils ont appris sur de nouvelles images. »
Usbek & Rica | 27 juin 2018

Design de l'attention

• Design et Attention : perspectives critiques

Pour l’enseignant chercheur en design Anthony Masure (@anthonymasure), la question de la conception attentionnelle doit être replacée en contexte : celui de la conception des interfaces.
Internet actu | 27 juin 2018