Daniel Mügge

Digital sovereignty is a big thing in Brussels - also in #AI. But what are the politics behind the "EU AI sovereignty" ambition?

After all, it could mean boosting EU AI at the expense of foreign competitors. But also - a completely different angle - to rein in tech influence in society tout court, so that citizens regain some sovereignty themselves. Which is it?

For the Journal of European Public Policy, I've dissected 6 yrs of Commission strategy to find out.

Now OA: https://www.tandfonline.com/doi/full/10.1080/13501763.2024.2318475

Students can pretty much automate take-home essay writing with natural language processing AI.

When I've tried it (e.g. sudowrite), auto-generated text easily beat average undergraduate essays, with no way to detect fraud.

Feels like there's a massive challenge coming to higher education, and universities are completely unprepared.

Whether we like it or not, the skills we teach, and more importantly how we test them, will have to change drastically in just a few years. This won't go away...

"The smarter we make the A.I., the less it wants to do our jobs."

The alignment problem summarized in the New Yorker of 31 Oct.

@frantecol@fediscience.org Dank voor het delen - inderdaad schrikbarend. Natuurlijk kun je de aandacht die er nu voor dit onderwerp is ook misbruiken. Maar in dit stuk wordt gewoon verondersteld dat dat de norm is, zonder enig bewijs. Terwijl machtsmisbruik, intimidatie, sexisme en racisme, etc. in de wetenschap wél uitgebreid gedocumenteerd zijn. Kortom, waar in vredesnaam baseert Levi dit op? Ook in dat opzicht volstrekt ongepast voor iemand in zijn positie.

Trade data is a staple of economic analysis. But, as this paper w/Lukas Linsi and Brian Burgoon shows, there is a lot more noise in the data than meets the eye, and than over-confident econometric analyses let on.

W/a new trade data set that reflects this noise (available here: https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/RBRHFE), we ran replications of prominent studies in the field - with some surprising results.

OA now in ISQ: https://academic.oup.com/isq/article/67/2/sqad020/7085502

Thanks to everyone who helped make this better over the years!

Sofia Scasserra and Adriana Foronda dissect how secrecy jurisdictions for tax evasion (Virgin Islands, etc.) can figure in corporate schemes to put data as a profit source beyond regulators' reach.

Similarities between tax havens and potential data havens go far:

- Leverage sovereignty to ensure minimum transparency
- Facilitate complex corporate structures to avoid regulation
- Strong infrastructure
- Limited ancillary regulation (e.g. climate impact of data centres)

https://www.tni.org/en/publication/banking-on-data

On newspaper front pages this morning: frustration about the lack of #COP27 ambition, which was curtailed by the world's largest fossil fuel producers.

And pictures from the opening game of a Qatari football world championship that is obscene in its wastefulness and saddening in its disregard for universal rights.

Can't help feeling that these are two sides of the same coin...

Last week, the EU reached political agreement on Digital Rights and Principles; final signatures expected in the coming weeks.

Often, skepticism in the face of lofty principles is justified. But in spite of the proposals limitations, I sense here a genuine ambition to make digital tech work for societies. If nothing else, European leaders signing up to that creates a standard to which we can hold them.

#GlassHalfFull this morning...

https://digital-strategy.ec.europa.eu/en/policies/digital-principles#tab_1

Who is lobbying whom about the impending EU AI rule book?

Hard to know from the outside, of course. But this report from the Corporate Europe Observatory shines at least some light on these efforts. The intensity of lobbying is as good a reminder as any about just how much is at stake here for big tech - and for the rest of society.

https://corporateeurope.org/en/2023/02/lobbying-ghost-machine

Getting ready for teaching about global ramifications of proliferating #AI, and once again amazed by the Anatomy of an AI System-map by @Katecrawford@mastodon.social and Vladan Joler. So much there, almost wall paper material...

Also remarkable how the whole regulatory infrastructure and the political context (public sponsorship, military use of AI, etc.) are not yet reflected here. Looks like there's another map, about the Political Entanglements of an AI System, still to be made ;)

https://anatomyof.ai/

Struggles over what privacy actually means hit a tech-saturated US campus.

Excellent story on the Mites project at Carnegie Mellon from the MIT Tech Review. Brings together abstract but fundamental debates about our future with a real-people, here-and-now narrative. Really insightful...

https://www.technologyreview.com/2023/04/03/1070665/cmu-university-privacy-battle-smart-building-sensors-mites/

@jensclasen@mastodon.social I just found myself trying to say "Na ja" with each of those meanings in front of my computer, with just slightly different inflections... Good that my office mate has already called it a day 🙃

@LarsJohannessen@sciences.social Beautiful topic. Too bad it's completely outside of my field 🙂

@phdskat@fediscience.org That sounds like a good start, I think.

@cqschneider@fediscience.org Agreed, it will take direct, personal interaction - what's the alternative? That's also much more labor intensive, though, and impossible to do blindly, which is something I appreciate about essays whose authors I cannot identify.

But I'm not a fan of testing and grading, anyway - seems to generate to much stress and get in the way of growth. Maybe short-circuiting essays will allow us to re-think the whole grading-business altogether? 🙂

Security and military logics encroach on EU digital tech regulation.

In the Journal of European Public Policy, I argue that point in general, and specifically for the EU's budding regulation of AI.

For better or worse, it makes little sense anymore to detach AI regulation debates from those about (in)security, and vice versa.

Available open access here:

https://www.tandfonline.com/doi/full/10.1080/13501763.2023.2171090

Job alert ✨

Assistant professor "AI and European Democracies" at the University of Amsterdam

- (Very short) tenure track
- DL 5 Feb 2023
- Embedded in the Humanities Faculty

https://vacatures.uva.nl/UvA/job/Assistant-Professor-%28UD%29-in-AI-and-European-democracies/761323802/

Zur Abwechslung mal auf Deutsch: Es sind natürlich schon ganze Bibliotheken zu Politik, Kapitalismus und Klimakrise vollgeschrieben worden. Aber dieses Diskussionspapier von Jens Beckert hat eine Menge Sachen nochmal gut auf den Punkt gebracht:

Verkaufte Zukunft. Dilemmata des globalen Kapitalismus in der Klimakrise.

Lesenswert und erhellend, wie ich finde, wenn auch nicht gerade erbaulich 😉

https://hdl.handle.net/21.11116/0000-000B-6790-E

@annehellmuthhaeuser@mastodon.world @politicalscience@a.gup.pe Thanks, Anne. Worth mentioning: the pdf of "Artificial Intelligence with Chinese Characteristics" is downloadable for free through the link you shared!