Spread the love

AI leaders are going to have a tough time ahead. Their technology and presentations might be jaw-dropping but the story doesn’t end there. Problems arise with security, consent, and the threats AI poses. We discussed on the possible threats that could arise with AI and also on why a leading actress is suing OpenAI over consent. Now, Google is facing backlash from the users over incorrect advices.

Google Gemini gives incorrect answers
Could google trust Reddit data?

Some people who started using Google AI Overview, asked some questions to it only to be given back with incorrect suggestions. Yeah, like it advises you to have “one rock per day” as they are important for your digestive health, quoting UC Berkeley scientists. Another incorrect answer was to apply non-toxic glue to make pizza to give the sauce extra tackiness and give it a unique flavour. It seems it picked up this answer from an 11-year-old Reddit post. Also, according to AI Overview, Barack Obama was the first Muslim president and that there were 17 presidents.

A Google spokesperson assured that most of the incorrect answers are uncommon and some are deliberately doctored and cannot be reproduced.

Something similar happened with Gemini, remember? The market value too dropped by $100 million when people started pointing out that the AI gave historically incorrect answers and generated images of people with colour as Nazi-era German soldiers.

Any of you in software development has likely encountered incorrect solutions suggested by our AI “colleagues”. The question is, how long do we have to wait until we get answers that are 100% correct?

Google AI Overview picks up wrong answers from Reddit
Google AI Overview picks up wrong answers from Reddit

This is why I told you in the beginning, the days are going to be a bit rough ahead. It is not so easy to control something that reads data from the online resources. This data would contain both right and wrong information. Maybe in the parallel universe, these erratic ones turn to be the bad robots and the ones that are accurate would be the heroes. The bad might control the military and the heroes must work with humans to save the world. Plot of every AI sci-fi movie. But will this happen in reality?

Update: Google has recently corrected its model.

You May Also Like

+ There are no comments

Add yours