In 1995, he became an assistant professor at Yale University before moving to the University of Oregon in 1998, where he became a full professor of theoretical physics and Director of the Institute of Theoretical Science. Hsu's research has focused on a number of areas in particle physics and cosmology, including phase transitions in the early universe, the ground state of quark matter at high density, black holes and quantum information, minimum length from quantum gravity, dark energy, and quantum foundations. In July 2012, Michigan State University named him vice president for research and graduate studies. At the time, Inside Higher Ed and Lansing State Journal described the appointment as controversial, due to Hsu's comments endorsing research into using genetic modification to increase human intelligence, and his blog posts describing human race categorization as biologically valid. On June 10, 2020, the MSU graduate student union began calling for Hsu to be removed from the administrative position. The MSU student association also called for his removal, and multiple petitions were circulated, including a counter-petition. As of June 17, petitions for removal had 700 and 470 signatures, while the counter petition had over 970 signatures. On June 19, 2020 MSU President Samuel L. Stanley announced that Hsu had resigned as Vice President, returning to a tenured faculty position. Hsu made clear that President Stanley requested his resignation, and that he did not agree with Stanley's decision.
Technology work
In 2000, Hsu went on leave from the University of Oregon to create Safeweb, an anonymizer service. In 2003, Symantec acquired SafeWeb for its SSL VPN technology. Hsu is a founder of Genomic Prediction, a company that develops technology for advanced genetic testing. Hsu also has an interest in psychometrics and human genetic variation, which he writes about in his blog and in other publications. In 2017, Hsu and five collaborators published a paper in Genetics on the use of lasso to construct genomic predictors of complex human traits, using data from the UK Biobank. Their genotype height predictor estimated adult height within an accuracy of roughly one inch. In 2018 his research group used the method on the same dataset to build genomic predictors for complex diseases such as Hypothyroidism, Hypertension, Type 1 and 2 Diabetes, Breast Cancer, Prostate Cancer, Testicular Cancer, Gallstones, Glaucoma, Gout, Atrial Fibrillation, High Cholesterol, Asthma, Basal Cell Carcinoma, Malignant Melanoma, and Heart Attack. Outliers in risk score were shown, in out-of-sample validation tests, to have up to ten times the risk of ordinary individuals for the specific conditions. The predictors use as input information dozens to thousands of common SNPs measured for each individual. He serves as scientific adviser to BGI, and as a member of its Cognitive Genomics Lab.
