Background Bicuspid aortic valve (BAV) is the most common congenital heart condition. BAV complications include aortic valve surgery, aortic surgery, infective endocarditis, and aortic dissection. Due to its 10% chance of transmission to offspring and lifetime morbidity burden of >80%, screening of first-degree relatives is recommended for early detection. Several barriers to echocardiographic screening for BAV have been identified, therefore, a cheap and fast “rule-out” BAV tool is appealing. The objective of our study is to create an artificial intelligence (AI)-enabled electrocardiogram (ECG) model to aid the detection of BAV in suspected individuals. Methods Between January 1, 1990, and June 30, 2023, 13,066 adults mean age 58.3±17 years, 46.6% women with both an ECG and an echocardiographic study within 6 months of each other, were identified from the Mayo Clinic database. Cases were selected as patients with a confirmed BAV diagnosis either by echocardiographic or pathologic report. Controls were selected as patients with an ECG indication of abnormal auscultation, rule out, suspected, or family history of BAV. Patients were excluded if they displayed any of the following criteria: a previous heart surgery or procedure, potential cardiac remodeling conditions(cardiomyopathies, history of chemotherapy or radiation, autoimmune disease, previous known congenital heart disease, heart failure, myxomas, history of endocarditis, rheumatic heart disease, heart transplant and pacemaker). ECGs by patient were randomly shuffled into mutually exclusive training (80%), internal validation (10%), and test (10%) data sets. A convolutional neural network was built using the Keras Framework under Tensorflow backend (Google;Mountain View,CA,USA). Results BAV prevalence in the model’s development was 45%. In the test group, the AI-enabled ECG model achieved an area under the curve (AUC) of 0.704 with the following values: sensitivity 65%, specificity 64.8%, accuracy 64.9%, positive predictive value (PPV) 60.1%, and negative predictive value (NPV) 69.4%. Conclusion An AI-enabled ECG model has shown modest diagnostic performance in detecting BAV. When assuming a prevalence of 10% (chance of transmission to offspring), the NPV increases to 94.3% with decreased PPV of 17%. This approach could save a major number of echocardiograms for “ruling-out” BAV, but needs external validation.