mdlincoln

Prolific is not just another player in the AI space – we are the architects of the human data infrastructure that's reshaping the landscape of AI development. In a world where foundational AI technologies are increasingly commoditized, it's the quality and diversity of human-generated data that truly differentiates products and models.

We’re looking for impact-focused Software Generalists to join our specialized team focused on serving frontier model creators and enterprise AI application developers. As a full-stack engineer, you will work across Prolific’s domains to solve customer and product problems.

This is an exciting opportunity to work directly with frontier AI companies, making critical technical decisions that balance scrappy startup execution with scalable, reliable engineering, as Prolific revolutionizes research for the AI community. You'll will have regular in person collaboration with customers and our US team, as well as collaborate closely with our UK-based tech teams.

Unfortunately we don't sponsor US visas at this time.

Apply: https://job-boards.eu.greenhouse.io/prolific/jobs/4767348101

Almost all the engineering problems cited in the original post are still basically there, but graphical models are still the least painful way of doing this, particularly when trying to share data between institutions. Example: https://linked.art/model/assertion/

>It takes a tremendous amount of work to make the work that goes into photographing this goblet invisible.

https://twitter.com/john_overholt/status/991110369082068992

(I work at the Getty Research Institute)

It results in code that more closely resembles executed order of operations (e.g. filter -> mutate -> group -> summarize). Context is also key: it's most often used for data processing pipelines in specific analytical scripts or literate-code documents - less so used when defining generalizable/testable functions in packages (again, just a personal perspective - YMMV of course)

For example, the Rembrandt Research Project infamously set out to identify genuine vs. fake Rembrandt paintings in his corpus of known works under the false assumption that there would be a lot of 18th/19th/20th century forgeries. In fact, most of the "non-Rembrandt" cases they found were not later imitations, but instead works done by his own students or contemporaries - or works co-produced by Rembrandt and another. The result - deconstructing the project's original false assumption - proved revolutionary for our understanding of artistic studio practice from the period, but failed to locate many "forgeries" as such.

A review (paywalled, sorry!): http://www.sciencedirect.com/science/article/pii/02604779899...

And a Met exhibition: http://www.metmuseum.org/art/metpublications/Rembrandt_Not_R...

https://github.com/NYPL-publicdomain/data-and-utilities

I should probably add that the types of "explanations" I put forward in this post are actually not of central concern to me - certainly not explanations derived solely from parsing quantitative results. I'm far more interested in the descriptive evidence this kind of measurement can provide. It can give wider context to what tends to be a very case-study-centric discipline (e.g. oh, this guy happened to work a lot with Italian publishers in this period? We didn't realize it before just looking at 5-10 artists per article/monograph, but actually that is quite exceptional/normal for this period...)

Then again, proposing these kinds of explanations is also something of a disciplinary norm, for better or worse.

A project apparently exploring how copyright claims result in selective censoring in the Street-View-esque images of museum collections produced by the Google Art Project.

The WaPo article actually conflates copyright and reproduction rights (I work in a museum curatorial dept. FYI) Copyright would apply to works where artists, or their estates, can still make copyright claims over their artworks (although the role of fair use in reproducing images of art is evolving: http://www.collegeart.org/fair-use/)

But why can older artworks that are now in the public domain still have their images blurred out? Although the museum may have agreed to openly release representations of the public domain works that they own, it is often the case that museums may temporarily hang works on loan from private collectors in their galleries. In cases like these, museums and the lenders work out loan terms that frequently include provisions about photography. These loan agreements supersede copyright issues. Whether or not museums should agree to such terms is, of course, a good question.