https://padlet.com/isabellanikolaidis/6q0y7ruqpeiyjws3 en-us 2020-11-03 00:46:14 UTC 2020-11-03 01:34:03 UTC hello@padlet.com isabellanikolaidis https://padlet.com/isabellanikolaidis/6q0y7ruqpeiyjws3/wish/884541575 2020-11-03 00:49:29 UTC https://padlet.com/isabellanikolaidis/6q0y7ruqpeiyjws3/wish/884541575 isabellanikolaidis https://padlet.com/isabellanikolaidis/6q0y7ruqpeiyjws3/wish/884543000 - relates both to the system and the societal context in which the system is deployed
- AI for Accessibility initiatives (nothing frontfacing about security, privacy)]]> 2020-11-03 00:50:19 UTC https://padlet.com/isabellanikolaidis/6q0y7ruqpeiyjws3/wish/884543000 isabellanikolaidis https://padlet.com/isabellanikolaidis/6q0y7ruqpeiyjws3/wish/884543665 2020-11-03 00:50:38 UTC https://padlet.com/isabellanikolaidis/6q0y7ruqpeiyjws3/wish/884543665 isabellanikolaidis https://padlet.com/isabellanikolaidis/6q0y7ruqpeiyjws3/wish/884543968 2020-11-03 00:50:49 UTC https://padlet.com/isabellanikolaidis/6q0y7ruqpeiyjws3/wish/884543968 isabellanikolaidis https://padlet.com/isabellanikolaidis/6q0y7ruqpeiyjws3/wish/884544524 2020-11-03 00:51:03 UTC https://padlet.com/isabellanikolaidis/6q0y7ruqpeiyjws3/wish/884544524 isabellanikolaidis https://padlet.com/isabellanikolaidis/6q0y7ruqpeiyjws3/wish/884547235 2020-11-03 00:52:31 UTC https://padlet.com/isabellanikolaidis/6q0y7ruqpeiyjws3/wish/884547235 https://padlet.com/isabellanikolaidis/6q0y7ruqpeiyjws3/wish/884549955 We are accountable for how our technology impacts the world.
Accountability - a structure that is put in place that we are enacting our principles
Part of Accountability is to help customers and partners be accountable - Have a set of principles on how we develop and sell and how we advocate for regulation on facial regulation - It has a lot of great uses, but a lot can interfere with social liberty.
Facial recognition principles - Take a high-level principle, and see how our development team at each stage is thinking about these at every life cycle.

When you look into resources, you see that Msft has "guidelines designed to help you anticipate and address potential issues throughout the software development lifecycle", but not the "guidelines they use to anticipate and address potential issues"

AI fairness checklist:
the most beneficial outcome of implementing an AI ethics checklist may be to prompt discussion and reflection that might otherwise not take place



]]> 2020-11-03 00:53:58 UTC https://padlet.com/isabellanikolaidis/6q0y7ruqpeiyjws3/wish/884549955 https://padlet.com/isabellanikolaidis/6q0y7ruqpeiyjws3/wish/884557814 Emphasis on the use of data to inform decisions, protecting that data is a key responsibility 

]]> 2020-11-03 00:58:14 UTC https://padlet.com/isabellanikolaidis/6q0y7ruqpeiyjws3/wish/884557814 https://padlet.com/isabellanikolaidis/6q0y7ruqpeiyjws3/wish/884560337 AI should be understandable for everyone.
Transparency help mitigating unfairness in machine learning. It helps gain trust from the customer. People who develop AI systems should be open to how, why they are using AI, and the limitation of their AI system. People should understand the behavior of the AI system. Applying interpretability and intelligibility in the AI system. ]]> 2020-11-03 00:59:34 UTC https://padlet.com/isabellanikolaidis/6q0y7ruqpeiyjws3/wish/884560337 https://padlet.com/isabellanikolaidis/6q0y7ruqpeiyjws3/wish/884563866 1. AI is a positive thing for the organisation, and they have the expertise to implement it. They are not going into negative sides, so as to not deter people away from these solutions. [For an average business user, AI is modelling or machine learning. If MSFT is in the picture helping you, it works as an incentive to move in that direction] Branding: "Pursuing computing advances to create intelligent machines that complement human reasoning to augment and enrich our experience and competencies."

2. Responsible approach to AI - They are taking into consideration how someone can be against it, so they are selling on "We are the good people, we have the strategies to cover the bases" rather than educating people on using ai responsibly

3. Facial Recognition - MSFT was one of the first to produce FR algorithms, they were approached by law enforcement and military, but MSFT said they would not sell to them based on their ethics and morals after Amzn and IBM. Companies seem to have turned this into a virtue-signalling marketing technique to show expertise. In reality, this could have been as a result of multiple reasons not limited to selling products to public organisations is not the most profitable way to go when you're developing future base solutions.

4. They critique current checklists because of the yes/no type of questions, by summarising it as the ultimate goal of a checklist is to prompt discussion and reflection. "We found that AI fairness efforts are often the result of ad-hoc processes, driven by passionate individual advocates."

5. They do not mention privacy concerns/ security/ ethical practices while referring to specific projects. They are only talked about on blog posts, leaving it at surface level information. They do not talk about any organisation being held responsible for understanding the implications of products that they want to develop.

]]> 2020-11-03 01:01:27 UTC https://padlet.com/isabellanikolaidis/6q0y7ruqpeiyjws3/wish/884563866