Information vortex
Pexels
Sometimes you need to go forward to go backward. “Forward to the Past” doesn’t have the same ring as Back to the Future, but that’s exactly what’s happening with advances in science, AI and machine learning. Whether it be distant history, artifacts and writings or more recent history, recordings and movements, artificial intelligence will be able to make seen the unseen and to recreate damaged works of art, destroyed European cities, lost societies and to sort through giant datasets in no time.
One issue is that these more anthropological and humanities-centric approaches to AI are in constant competition with the funding and attention that tends to be attracted to some of the more futuristic applications of AI. With that said, there are still a number of exciting new projects and tools that have made getting a fuller picture of history and a more engaged civic body their priority.
Distant History
Want to learn from the distant past? You’ll probably need to digitize a bunch of old texts. LayoutParser, a new tool used by researchers at some of the most esteemed universities, intelligently scans passages and stores them in functional data structures. It also contains a robust tutorial side, assisting researchers in more ways than one.
GlobalXplorer is a new satellite imaging system that explores and catalogs artifacts, keeping a record of history in a place that no one can tamper with it.
Odeuropa is an AI that helps to recreate smells from history. This sensory experience of the past helps with immersion, simulation and getting to know a place better than ever before.
It’s getting easier to assemble and interrogate monstrous data sets—unimaginable just 36 months ago. Like other sciences, history is increasingly approached as a partnership with data scientists.
“Whether it’s new DNA studies where we’re able to understand not just that, okay, this group of mummies that was excavated in Egypt 100 years ago, we know that they’re royal. Now we know how they are related to each other,” said archeologist Sarah Parcak. “I think, with our own application of physics and chemistry and biology and computation machine learning — putting all these new tools to use, looking at the past, we’re far better able to understand all of their diverse technologies.”
Recent History
Cortico.ai is a new tool out of MIT that synthesizes conversation and creates a word map in real time to indicate various key topics throughout a conversation. It has the added ability to download individual snippets of conversation with ease, transcribe and annotate the transcript and much more. Coritco is interrogating transcripts of 100 Days of Conversation going on right now in almost every state.
Deep Trust Alliance: The first of its kind global network of stakeholders who are combining to fight AI with AI.
BLMProtestBot: Scans an image for where faces are and covers the face with a Black Lives Matter fist, protecting the identity of protestors to prevent additional police intervention.
Blackbird: A deepfakes/misinformation AI to help ensure that we seek and find the truth.
Citizen Evidence Lab: A crowdsourcing and machine learning tool for verification of citizenship through Amnesty International.
Implications
Tyler Cowen noted to archeologist Parcak, “Your job is almost becoming impossible.” History and the social sciences are now team sports requiring knowledge of languages and cultures, satellites and sensors, math and data science.
“With greater accuracy and precision towards the past we will not only be able to learn more about our collective history, but we will also be able to fill in the gaps with regards to bias and more. In history, among other places, assumptions can be greatly detrimental to the work and the key takeaways,” said Parcak.
These advances won’t be without controversy. Already numerous institutions are dubious of the work that these AI’s are doing — for both historical curation and civic conversations we will need a degree of transparency that is not yet normalized amongst technological innovations.
For high schools and colleges, it is clear that learners need more time and support to investigate big questions in interdisciplinary projects using machine learning tools. Beginning in middle grades, learners need the opportunity to investigate the ethics and economics of life with smart machines.
For faculty, the big implication is the increasing need to say, “I don’t know, how might we…?”
