Easier and Quicker Yes; More Reliable: No!
The inaccuracies in database research used to be just from input error, GIGO! (“garbage in, garbage out”). AI has, however “evolved”. AI makes decisions regarding the information without using human perception and experience, so mistakes occur
Here are a few serious examples of AI Hallucinations:
Criminal charges based on false AI: According to news articles, The UK Post Office relied on their accounting programme for many years, and found 700 managers guilty of theft or fraud. Some went to prison, others had to pay large sums of money. Unfortunately the programme, (accidentally or on purpose is not known at this time), created false shortfalls.
Fake AI case precedents in court: A problem in using AI database searches in the USA has nearly ruined more than one lawyer’s career. One lawyer used a well known AI program to prepare a filing, but the AI bot produced fake cases supporting his client’s position that the lawyer presented to the court. The lawyer was uneasy and wanted precedent, so the AI produced fake precedent. This is one of the initial cases in which AI “hallucinations” made it into court.
AI Voice mimicry: The National Post reported that a man tricked eight people into giving up nearly $200,000.00 to AI mimicked voices of their grandchildren. The CBC reported that the fraudster collected information about some of the grandchildren and used it to create a more convincing profile.
Do IBIS investigators use databases? Of course!, but we double source, using HUMINT, (human intelligence, that is a live human being investigating on the ground). Investigators are even more crucial in any Due Diligence file because the work of an AI database researcher has to be reviewed and verified by an investigator!