The Role of context in assessing AI Risk

The Role of context in assessing AI Risk

First, I must disclose that this newsletter will be super light. It is 11.45pm here is Prague and after three beers (or 4 if you count the welcome drink!), it is hard to write. Also, after a looong (yes, that is the correct spelling when you want to emphasize something very long!) red-eye flight from Boston and after a 10hr layover in Reykjavik, I must admit, I would love some sleep but let me share what I learnt today and why I am motivated to share it with you today rather than wait for tomorrow!

I am here in Prague attending the IEEE Robotics and Automation summer school on Multi-robot systems and had the amazing fortune of meeting some super smart researchers, students and scientists working on cutting edge products deploying AI, ML and autonomous systems. Robotics is challenging especially when you have to factor physical environments, mechanics and other constraints, aspects of control and planning etc. etc.!

Needless to say, the algorithms are complex and not all plans are destined to succeed in deployment. While this is an emerging and fascinating area of research, there is also an increased awareness of responsible AI in the field especially with respect to following laws(for UAVs for example), safety( working in places where humans could get hurt), guarantees, the question of automation and how much of activity could be autonomous.

In addition robots have to work in situations where they may not have access to a centralized server, communication networks may be constrained/not available. GPS/GNSS may not be available. environments with noise/potential collisions or lack of reference markers makes it challenging to accomplish the goals robots are designed to do.

"It just works - most of the time"

While some amazing capabilities in terms of factoring changing dynamics in environments, it is impossible to factor everything when you design robots and associated algorithms. While it would be fascinating to have an all-purpose, omnipotent platform that could do "anything", it is still far away from realization. So how do we take advantage of the state-of-the-art and build systems ?

The answer is design with the context in mind and assess it for the use case.

The importance of designing with a use-case in mind and the importance of factoring context when assessing a system for its suitability and capability to be used for the application is paramount to assess whether a system is ready for use!

In the next few newsletters this week, we will expand, through illustrations, how context drives the requirements when designing algorithmic systems!

Before I end this newsletter, coming back to beer, the workshop started today with organizer saying:

Starting from lunch, we will have beer!

He kept up his word, as the day ends with many samples of beers from the Prague region.

Here is one for memories and if you have never been to Prague, to inspire you to visit! Cheers!

No alt text provided for this image

??Keep on learning!

?? Want to learn more? QuantUniversity is offering a LIVE Algorithmic Auditing course on August 22nd 9.30 AM to 4.00pm in partnership with PRMIA. If you are interested, check out details here:?https://prmia.org/Shared_Content/Events/PRMIA_Event_Display.aspx?EventKey=8906

??Many of these topics will be elaborated in the?AI Risk Management ?Book published by Wiley. Check updates here ->?https://lnkd.in/gAcUPf_m

??Subscribe to this newsletter/share it with your network ->?https://www.dhirubhai.net/newsletters/ai-risk-management-newsletter-6951868127286636544/

I am constantly learning too :) Please share your feedback and reach out if you have any interesting product news, updates or requests so we can add it to our pipeline.

Sri Krishnamurthy?

QuantUniversity

#machinelearning ?#airiskmgt ?#ai

要查看或添加评论,请登录

社区洞察

其他会员也浏览了