Bench & Bar March/April 2026

ENDNOTES 1 Ben Schamisso, New AI Transforms Radiol ogy with Speed, Accuracy Never Seen Be fore, Northwestern University Engineering News, Jun 6, 2025, https://www.mccormick. northwestern.edu/news/articles/2025/06/ new-ai-transforms-radiology-with-speed-ac curacy-never-seen-before/ (accessed 2/2/2026). 2 Yusuf Ahmad , AI Is a Tool for Radiology, Not a Replacement for the Profession, American Col lege of Radiology, October, 2023, https://www. acr.org/News-and-Publications/AI-Is-a-Tool for-Radiology (accessed 2/2/2026). 3 Jing, A.B., Garg, N., Zhang, J. et al. AI solutions to the radiology workforce shortage. npj Health Syst. 2, 20 (2025). https://doi.org/10.1038/ s44401-025-00023-6. 4 Robert N. Charette, Michigan’s MiDAS Unem ployment System: Algorithm Alchemy Created Lead, Not Gold , IEEE Spectrum (Jan. 24, 2018), https://spectrum.ieee.org/michigans-mi das-unemployment-system-algorithm-alc he my-that-created-lead-not-gold [ https://perma. cc/CP8Q-PSUN] . 5 Cahoo v. SAS Analytics Inc. , 912 F.3d 887, 892– 93 (6 th Cir. 2019). at 893. 6 Cahoo, 912 F.3d at 892. 7 Royal Commission into the Robodebt Scheme vol. 1, v, Commw. of Austl. (July 7, 2023), https://perma.cc/3B7G-HUJJ, (accessed [here inafter The Robodebt Scheme] (accessed 6 Jan uary 2026). 8 Id. at xxv. 9 Id. 10 Id. at xxix. 11 Julia Angwin et al., Machine Bias , ProPublica (May 23, 2016), [https://perma.cc/6GUV-GQ ZJ]. 12 Furkan Gursoy, Ioannis A. Kakadiaris F. Gur soy and I. A. Kakadiaris,” Equal Confusion Fairness: Measuring Group-Based Disparities in Automated Decision Systems,” 2022 IEEE International Conference on Data Mining Workshops (ICDMW), Orlando, FL, USA, 2022, pp. 137-146. https://doi.org/10.1109/ ICDMW58026.2022.00027 Computational Biomedicine Lab. 13 See https://hivehub.org/ (accessed 2/2/2026). and information assurance in the computer engineering and justice administration disciplines. His focus is on law and social sciences as they relate to computer engi neering, evidence and digital forensics. Courses include digital and computer crime, transnational cybercrime and legal issues with data mining and infor mation assurance. He holds a J.D. and a B.S. in mathematics from Louisiana State University. ABOUT THE AUTHOR MICHAEL LOSAVIO teaches in the Department of Criminal Justice and the Department of Computer Engineering and Computer Science at the University of Louisville on issues of law, society

subject to financial penalties. 4 The penalties went beyond the recovery of any monies paid. A fine of up to four times the amount of claimed benefits was also levied and sent to collection against out-of-work claimants, leading to bankruptcies, evictions, and other misfortunes. 5 Later evaluation of the machine system found an error rate in fraud positive determinations of 93%. 6 Australia implemented an algorithmic system to assess eligibility for social benefits that was designed to recover overpayments to recipients going back several years and was expected to save 1.7 billion Australian Dollars over five years. 7 It would ultimately remove human involvement in the deter mination process for recovery of supposed overpayments. 8 Yet despite in-house counsel and policy analysts finding the system didn’t accord with the law and noted the risk of miscalculations, the government pro ceeded anyway. 9 After injuring thousands of people through errors, the government finally repaid 746 million Australian Dollars to those erroneously charged and wrote off and reduced to zero 1.751 billion Australian Dollars in claimed overpayments, settling the class action on the day of trial. 10 Algorithmic systems are now used in criminal justice to make decisions on risk of recidivism, punishment and placement. This can help, as with the U.S. Sentencing Guidelines, to reduce disparities across courts in the country. But issues remain. Angwin and Larson of the non-profit jour nalism project ProPublica found issues with racial disparities and accuracy in predicting recidivism; Furkan et al. of the University of Houston’s Computational Biomedicine Laboratory suggested it produced unfair outcomes due to gender and race, noting All is not lost, though, when thoughtful applications use the Human element to keep things straight. That begins with education on the proper use and wrongful abuse of AI. Rachel Singel, an art professor at the Uni versity of Louisville, looks at the challenge from a “…model that grounds AI in Art History and ethics, using the printing press as a historical parallel to show that while its use was “worrisome.” 11,12 BEING PREPARED

technology once distributed human ideas, its new ability to generate them requires stu dents to use their “Humanities-brain” as the machine’s essential director.” The Human in-the-Loop principle is vital. Rachel Hopp, a biology professor at Louisville, uses AI in biology classes to create new lessons and ease WCAG2 Web Content Accessibility Guidelines workflow for creating alt-text on images of anatomical subject matter. George Higgins of criminal justice began a class on fundamental literacy with AI systems with and focus prompt engineer ing and the importance of the human in the loop, reinforcing the importance of the human element and empowering stu dents with the agency to do the job right. And his department began a minor in AI applications, with classes examining ethics, programming and, yes, the legal and social issues with AI. And Adel Elmaghraby of computer science built the HIVE AI Innovation Hub out of contracts and gifts to strengthen competent and safe development of these systems. 13 These and the efforts of others across the Commonwealth, and our United States, and brethren world-wide will help us all best benefit from these amazing new technologies IN CLOSING The future is very bright, and also very dangerous. We push computing systems to market and thereafter look at the full scope of damage they might do. The Human in the loop is essential to minimize that. The United States does not have a federal AI Act, in part due to concerns of hindering innovation. Especially in our competi tion with the People’s Republic of China. The European Union does have an AI Act, which can inform us, but they, too, are con sidering revisions to promote innovation and competition. And this is, again, where your work as a Kentucky lawyer will be essential. The Rule of Law still applies, even to rich tech bros. Its enforcement is a key component of pro tecting our sister and brother citizens from these new challenges and maximizing the benefits for all Kentuckians.

39 bench & bar