Web
Analytics
DL Seminar | Untangling the Loop –Four Legal Approaches to Human Oversight
top of page
  • Writer's pictureDigital Life Initiative

DL Seminar | Untangling the Loop –Four Legal Approaches to Human Oversight


Reflection by Ka Wing Lui (scroll below).



By Ka Wing Lui

Cornell Tech


In today’s seminar, Professor Jake Goldenfein gave us a talk titled Untangling the Loop –

Four Legal Approaches to Human Oversight. Professor Jake Goldenfein is a senior lecturer at Melbourne Law School and has been a researcher at Cornell Tech, Cornell University. In the seminar, he discussed about four legal forms of human oversight for AI and automation.

Professor Goldenfein started with talking about two typical diagnoses of why human oversight fails, one diagnosis is that the model of the human is wrong, that more time and training are needed to make decisions, and more specifically, means to look to human factors for empirical truth of the human. Another diagnosis is that the law is too vague, and needs to provide more principles as scaffolding, and more specifically Sui generis laws for each technical application.


Professor Goldenfein then discussed about several related paradigms and how they

evolved overtime. Firstly, humans as a moral agent that make “accountable decisions”, which

entails that these decisions can only be made by humans. Then, human as a flawed operator for “distribution of responsibility”. Human computation capacity cannot be quantified. Not only dealing with human and machine, but always need to consider organizational context. Professor Goldenfein used deadly crash accident with self-driving Uber as an example. The next paradigm Professor Goldenfein mentioned is human as a safety feature with “product liability”. Now, human oversight became to serve as a service. For example, human loop to be purchased for AWS, which is to build and refine AI models with humans in the loop workflow. The last paradigm is that humans as rights attractors and creative labor, for example, the discussion that for a writing piece, is that only the portion written by humans can have copyright.


Overall, Professor Goldenfein's seminar serves as a compelling lens through which to examine the multifaceted interplay between human agency, technological advancements, and legal frameworks within the expansive landscape of digital life. By navigating these complex intersections, the discourse not only enriches our understanding of contemporary challenges but also paves the way for informed and ethical engagement with emerging technologies.

bottom of page