(1.) A computer-driven process for informing the loan applicant very quickly, sometimes within a few minutes, whether the applicant will be approved, or whether the application will be forwarded to an underwriter. The quick decision is based on information provided by the applicant, which is subject to later verification, and other information retrieved electronically including information about the borrower's credit history and the subject property.

(2.) A computer-generated loan underwriting decision. Using completed loan application information, an automated underwriting system retrieves relevant data, such as a borrower’s credit history, and arrives at a logic-based loan decision. Some applications may be referred to manual underwriting, and some inputs, such as income and assets, must be verified at a later date.

Bankapedia's Take

While the merits of automated underwriting are undeniable, so are its drawbacks. The days of going into a bank sitting down with the person that will ultimately be given authority to decide yeah or nay on your loan are long gone, and largely in part because of AU systems. While an Automated Underwriting doesn't care what race or color you are, it also isn't always able to make common sense decisions that a human might be able to make fairly easily. For example if a person is near retirement age, and tells you they plan on retiring and living off of a fixed income in a few years, and they are applying for an ARM loan. Future employment is something an AU system simply can not discern. And since loan officers end up taking little to no responsibility for a failed loan, they are happy to let the AU system spit back a yes, only for the loan to ultimately fail. 

AU systems and the complex algorithms that run them, can help in mitigating a banks risk, but relying to heavily on numbers, which these systems effectively enable, can really end up biting a bank. The mortgage crisis we are currently undergoing is evidence of this. How often does one hear about a congressman, who has no experience in finance ask how so and so could possibly get a loan, and the fact that banks let computers make the bulk of their decisions certainly factors into the answer.

Referencesautoma

Guttentag, Jack. "Mortgage Glossary". 2008 <http://www.mtgprofessor.com/glossary.htm>.

"answers.com". 2008 <http://www.answers.com/Automated%20Underwriting>.