During the presentation Google I/O 2024 In addition to new artificial intelligence technologies, the software giant unwittingly demonstrated the main drawback of modern AI — mistakes, incorrect and harmful advice.
Google’s presentation «Search in the Gemini era» showed a new feature — video search. An example was a video of a stuck film advance lever on a film camera, accompanied by a question: «Why doesn’t the lever move all the way?»
The AI responded by offering a very «interesting» solution to the problem, which could destroy the footage.
So, the language model suggested «light» film. Anyone who has used a film camera knows that this is a terrible idea. If you open the camera while you’re shooting, you can ruin some or even all of the film because of the light that hits it. It’s literally the worst thing to do in this situation.
It’s also worth mentioning that this is not the first time Google’s language models have been caught distorting facts. Last year, the chatbot Bard claimed that the space telescope «James Webb» was the first to photograph a planet outside the solar system (this is not true). Earlier, Google employees called Bard «a pathological liar».
And the new Google Gemini AI was «caught» distorting historical data — in images it refused to reproduce white people where they should have been in the context.
Source: The Verge
Spelling error report
The following text will be sent to our editors: