Abstract
There is a lot of buzz recently about retrieval augmented generation, which refers to generative artificial intelligence that first generates a response and then also taps into search engines to confirm or update results. Generative artificial intelligence involves an algorithm training on large amounts of data, finding patterns in the data, and then using those patterns to make a prediction. For textual uses, an application which has been popular for a couple of years now is the algorithm looking at lots of text files to train, then after training being able to provide a textual response to a prompt - a chatbot. The two benefits of retrieval augmented generation are: it can help to curb misinformation (1) coming out of generative artificial intelligence, and, if an algorithm was trained on data in the past with a cutoff date for training, combining the trained algorithm with a continuously updated search engine can be a way to bring in more recent information or changes since training into the chatbot’s output. This article covers some of the implications of retrieval augmented generation, along with some of the economic and legal background, which might tend to incentivize incorporating retrieval augmented generation into chatbots (or to incentivize traditional search engines to incorporate chatbots, as the case may be).
Original language | American English |
---|---|
Pages (from-to) | 5-6 |
Number of pages | 2 |
Journal | Technical Services Law Librarian |
Volume | 49 |
Issue number | 4 |
State | Published - Jun 1 2024 |