Here’s some food for thought (literally) when weighing the pros and cons of the rapid development of AI in such areas as – well I guess pretty much everything. Kasia Chmielinski, from the Data Nutrition Project – an organization that researches challenges of ethics and governance of Artificial Intelligence – has come up with a clever metaphor (that involves sandwiches) to make a case for the development of more transparent algorithms. Chmielinski gives examples from her experience as an AI develop/researcher to illustrate that we too willingly accept and use AI products without knowing anything about the data that was used to build them. (This reminds me of a comic who noted that he would refuse to try a new food that his own mother would offer him, but would pop a free sample from a stranger in Costco without a second thought...) After pointing out the major concerns and challenges about AI data knowledge and transparency gaps, Chmielinski reports on the recent progress regulators and organizations such as Data Nutrition have been making (good, but lot more to do), and outlines some basic principles that need to be applied to make AI use safer and fairer. Knowing how the AI sausage was made and what it’s made of would be very reassuring. (I’m still going to accept any and all free food samples that come my way though...) #AI #modernIT
Kasia Chmielinski: Why AI needs a "nutrition label"
https://www.ted.com
Jens Nestel
8mo
Recipe for transparency lacking vital ingredients?
Compound Learning Expert | AI + Second Brain | Turning Knowledge into Power | Ex-Blinkist, Ex-Idealo, Ex-Zooplus
8moThanks for sharing Olga Selina. This data labeling system sounds like a great step to ensure a higher data input quality for AI models.