Slovakian startup CulturePulse is a social media analytics company and the UN is hoping it can use AI to better understand the Israeli-Palestinian conflict.
CulturePulse uses large datasets and machine learning to build behavioral and psychological digital twins of audiences. Marketing companies can test their advertising on these digital twins to see if their messaging will have the desired effect.
CulturePulse thinks that this same principle can be applied to model conflict zones and help organizations like the UN make better decisions in their attempts to resolve them.
The company says its product models belief systems that can “quantify anger, anxiety, personality, morality, family, friends, finances, inclusivity, racism, hate speech,” and more than 80 other social categories.
The UN engaged the company in August for a 5-month trial project during which CulturePulse will create a digital twin of the entire conflict area and its people. The recent escalation in the region has heightened the project’s expectations.
Essentially the company will create a kind of SIM City AI model with digital twins of every one of the 15 million people who live in Israel and the Palestinian territories.
With an accurate model, you could introduce a resolution approach and simulate how it would play out. You could ask, “If we did this, how would people react?” and then see if the decision would likely have positive outcomes or unintended consequences.
Will it work?
CulturePulse claims its model predicts real-world outcomes with a 95% accuracy rate. But does modeling an audience for marketing translate effectively in a scenario where people are killing each other?
CulturePulse CEO Dr. Justin Lane thinks it does. His team used their AI model to analyze other conflicts in the past, notably the troubles in Northern Ireland. The model accurately predicted the escalations there in response to Brexit in 2019.
AI has advanced a lot since then, so it may be that their model is a lot more accurate now than it was back then.
To model an extremely complex situation like the Israeli-Palestinian conflict requires massive amounts of data. The source of that data is another interesting project that AI and machine learning have made possible.
🌎 A recent uptick in our Conflict Tracker spotlighted the ongoing situation in Israel and Palestine, stirring deep reflections within our team.
These aren’t mere data points. They represent real lives, stories, and struggles happening on the ground, which our CEO, Justin E.… pic.twitter.com/aevIrznLqP
— CulturePulse (@culturepulseai) October 9, 2023
The GDELT Project
The CulturePulse AI model relies on data extracted from more than 50 million news articles from The GDELT Project.
GDELT uses natural language and data mining algorithms to monitor nearly the entire world’s news media and “extract more than 300 categories of events, millions of themes and thousands of emotions and the networks that tie them together.”
The idea behind the project is that if you can grab every piece of the world’s broadcast, print, and web news then you get a real-time snapshot of human society. Using AI makes it possible to parse, translate, and label the content so it can used by companies like CulturePulse.
By focusing on data related to the region in question, CulturePulse believes it can accurately model the emotions, beliefs, thinking, and anticipated actions of people on both sides of the Israeli-Palestinian conflict.
They also add the AI-generated disinformation that has fueled the conflict to their dataset to model the influences that people are being exposed to.
Will an AI model be able to make sense of a situation that humans have been unable to resolve for decades? Considering the results from initiatives like AI-powered predictive policing, it doesn’t look promising. We’ll have to see if the UN renews the project in January.
If nothing else, projects like these are teaching AI models a lot about humanity, for better or for worse.