One of the things I research is the cooperation of cultural institutions like museums in pitching tech products--specifically AI--to the public. This is advertising for the capacity of AI to educate and thus to reshape education itself as a project better left to private tech companies. 1/n
“The installation, unveiled in February 2024, allows guests to ask Douglass questions through an iPad. Using a closed database of Douglass’s autobiographies and speeches, the artificial intelligence generates responses based solely on Douglass’s recorded words.”
www.bostonglobe.com
In this case we have a public-private partnership between "MA250, the state [org] responsible for celebrating the 250th anniversary of the signing of the Declaration of Independence" & Timelooper "Interpretive Experience Design firm" per their website along w participation of other start-ups. 2/n It doesn't really matter here that this is profoundly corny. Seriously, look at this image. Also, amusing because Douglass was reputedly the most photographed individual of the 19th century who left behind a mass of writing. But this is almost beside the point for reasons I'll lay out. 3/n
What's important isn't the quality. It's the sales pitch for AI products as having not simply the *capacity* to educate, but the *right* to be used in this way, which, as I type this out, is now a mandatory part of secondary school education in Boston. forbes.com 4/n When I was doing my PhD I worked on the side as a historical guide in lower Manhattan. The experience was dialogic & vibrant & messy & certainly not the sanitized, deceptively "immersive" play-acting that is being offered by these companies. But again, quality is irrelevant. This is marketing. 5/n So it's normalizing AI as a legit educational aid. That's the point, and that's my main point too. But I do have one more thing to add, and it's this. The president of the Museum of African American History said the following to allay some concerns. This is a common defense. 6/n
The defense is: "we use a vetted set of authorized texts for the purposes of training AI to generate responses to questions." Aside from the uncanny valley horror of necromancy that such automated history-telling incites, there's also the issue of sources being processed like this w/o context. 7/n Let me give you an example from a different context. Elie Wiesel has recently been AI-conjured for similar purposes in the classroom. But it's important to think abt the nature of the sources being used: that is, Wiesel in translation. Why is this a problem? 8/n bloomberg.com
www.bloomberg.com
This is a problem because as Naomi Seidman addressed (Faithful Renderings), the language of Wiesel's revenge fantasies in the original was radically toned down in English translation to create a more docile narrator. 9/n There's an enormous difference between reading and teaching the toned-down book and AI-generating a persona to answer student queries based on that neutralized archive. And that's just one problem with this form of historical fantasy-spinning. 10/n 40 years ago Langdon Winner & David Noble warned that critiques of tech in society need to be about BOTH the social & political context of tech’s promotion AND the functionalities of tech itself. Focus on only the former leads to faulty social constructivism & the latter to tech determinism. 11/n So what I am saying here is that just claiming that AI is mid misses the political and economic project it’s being leveraged to bring about (upward wealth transfer), while only focusing on that project without noting the flaws of the tech risks missing HOW it’s carrying out that project. 12/n