\n
\n
\n Media contact:<\/strong> Brittany Steff, bsteff@purdue.edu<\/a> <\/p>\n <\/div>\n
WEST LAFAYETTE, Ind. \u2014 A coin flip. A roll of the dice. A shake of the Magic 8 Ball. Humans have been outsourcing their decision-making to inanimate objects since long before recorded history.<\/p>\n\n\n\n
But when a Magic 8 Ball reports \u201coutlook good,\u201d it\u2019s random chance. It lacks any information about the situation. Alex Psomas<\/a>, assistant professor of computer science in Purdue University\u2019s College of Science<\/a>, is designing artificial intelligence models to outsource decisions to make better, more informed and fairer choices, including how best to distribute food among food pantries in Indianapolis.<\/p>\n\n\n\n The process takes all the emotion out of the equation, something humans find impossible to do, and harnesses innovative AI models to weigh all the possible considerations and come up with the solution that best benefits the most people.<\/p>\n\n\n\n Psomas is teaching dispassionate computers to help humans make weighted \u2014 and weighty \u2014 decisions, including how best to allocate food pantry resources. It\u2019s not just a question of deciding who \u201cdeserves\u201d an apple between two people. It is quantifying how many people have what degree of need and how much each resource can help them and improve their lives.<\/p>\n\n\n\n \u201cEconomists think about value as being quantitative,\u201d Psomas said. \u201cThe utility I have for an object, like a phone, or like a carton of food, depends on my priorities, my circumstances and my values. Any calculation has to take into account everyone\u2019s viewpoints and make everyone happy with the outcome so that it feels fair to everyone. There are lots of different ways to define fairness, but our goal is no envy. That\u2019s the gold standard. We want everyone to be happier with what they get versus what someone else got.\u201d <\/p>\n\n\n\n Technically, flipping a coin, the simplest possible decision-making aid, is binary. That binary code forms the basis of how computers think, but the process has advanced light-years in the last century, and even in the last few years.<\/p>\n\n\n\n \u201cI really care about big applications, but, at the same time, I\u2019m a theoretician,\u201d Psomas said. \u201cI just do a bunch of math all day, and for me one of the coolest things about this project is that, not only are the outcomes beneficial, but the math is actually pretty cool. The math we\u2019re using is based on a set of ideas that were revolutionary in computer science.\u201d<\/p>\n\n\n\n The revolution is what is referred to as \u201cload balancing.\u201d Load balancing techniques are used to regulate situations including distributing network traffic in a router to be sure that everyone gets the best possible speed.<\/p>\n\n\n\n Psomas\u2019 background is in computer science and logic. He has worked extensively with systems where he uses numbers to model nebulous ideals, including incentives, fairness and value.<\/p>\n\n\n\n \u201cI am very interested in problems in economics,\u201d Psomas said \u201cMy strength is approaching economics problems from a computer science angle and approaching computer science problems from an economic angle.\u201d<\/p>\n\n\n\n The fields of economics and mathematics, especially the math used in AI models, have a great deal in common, but not many researchers are conversant in both worlds as is Psomas. Potential applications for this kind of work include new possibilities for solving societal struggles, including regulating organ donation, incentivizing preventative health care and distributing water for fair use. As part of the Purdue Computes initiative<\/a>, Psomas works to advance the frontiers of computer science and AI.<\/p>\n\n\n\n \u201cThere are tons of a traditional optimization problems, but with fairness factored in,\u201d Psomas said. \u201cThey are very fascinating problems, and the solutions can save lives.\u201d<\/p>\n\n\n\n One of his more recent, and dynamic, projects included a more efficient way to distribute food donations to food banks<\/a> in partnership with the Indy Hunger Network<\/a>.<\/p>\n\n\n\n The Indy Hunger Network has an initiative, known as Food Drop, that allows shipments of edible food to be efficiently donated to food banks. Food may be rejected from a grocery store, restaurant or other consumer because of logistical errors or imperfections, but the food is still perfectly edible. Traditionally, that food would be bound for the landfill. Food Drop instead allows this food to be donated to a food bank.<\/p>\n\n\n\n \u201cThey would call around to food banks and figure out who could take the load of food \u2014 who needed it the most. But those calls took resources. We worked with them to automate the system to figure out, automatically, where the food should go.\u201d<\/p>\n\n\n\n Psomas\u2019 innovative machine learning program saves time and energy<\/a>, making automatic allocations rather than just recommendations. The Indy Hunger Network staff no longer has to call around to food banks, which requires the food banks to answer the phone and be able to instantly relay information. And the truck drivers no longer have to idle, waiting to be given a destination. He talks more about the program in this video<\/a>.<\/p>\n\n\n\n This work, joint with Psomas\u2019 graduate students Marios Mertzanidis and Paritosh Verma, appeared in the Association for Computing Machinery<\/a>\u2019s Conference on Economics and Computation 2024<\/a> and recently won the award for best student paper at the Association for Computing Machinery\u2019s Conference on Equity and Access in Algorithms, Mechanisms and Optimization 2024<\/a>.<\/p>\n\n\n\n \u201cThis is one of the largest-scale systems that we know of that uses AI to help food distribution and explicitly models and addresses considerations of fairness,\u201d Psomas said. \u201cWe built it, we automated it, we maintained it, and now we\u2019ve handed it off to the Indy Hunger Network to use. It\u2019s open source: The code is out there for anyone to use. I hope this grows beyond Indiana. The computers can free up the humans to address other problems, and everyone gets fed. It\u2019s a win-win.\u201d<\/p>\n\n\n\n Psomas\u2019 research is funded by the National Science Foundation<\/a>, as well as a Google AI for Social Good<\/a> award.<\/p>\n\n\n\n Purdue University is a public research institution demonstrating excellence at scale. Ranked among top 10 public universities and with two colleges in the top four in the United States, Purdue discovers and disseminates knowledge with a quality and at a scale second to none. More than 105,000 students study at Purdue across modalities and locations, including nearly 50,000 in person on the West Lafayette campus. Committed to affordability and accessibility, Purdue\u2019s main campus has frozen tuition 13 years in a row. See how Purdue never stops in the persistent pursuit of the next giant leap \u2014 including its first comprehensive urban campus in Indianapolis, the Mitch Daniels School of Business, Purdue Computes and the One Health initiative \u2014 at https:\/\/www.purdue.edu\/president\/strategic-initiatives<\/a>.<\/p>\n\n\n\n Podcast Ep. 128: How Purdue Is Using AI for Good \u2014 Computer Science Professor Alex Psomas Explains<\/a><\/p>\n\n\n \n Media contact:<\/strong> Brittany Steff, bsteff@purdue.edu<\/a> <\/p>\n <\/div>\n Fair and square<\/h2>\n\n\n\n
Thought for food<\/h2>\n\n\n\n
About Purdue University<\/h2>\n\n\n\n
Podcast<\/h2>\n\n\n\n
\n \n \n