Discussions on AI’s overall problems took place online, proving a difficult start for the transition of the Google research AI. In response, the technology giant published a statement on May 30, 2024.
The company explained that most of the “Viral IA” stories were false. The overall errors of the real AIs were rare and generally came from common problems of the search engine.
READ: How to deactivate Google Ape
The article continues after this advertisement
However, Google has announced that it would implement more strict guarantees and more closely monitor its new feature to minimize errors.
What caused AI’s overall problems?
AI previews summarize data from the best online searches and place it at the start of online search pages. Consequently, users save time when you are looking for relevant information without having to go to several websites.
It also applies generative AI technology from Chatppt and programs similar to online searches. Unlike popular belief, they do not use language models alone.
The article continues after this advertisement
The company has integrated it into its main web classification systems to perform “traditional research tasks”. These include identifying the high quality results of the Google index.
The technological company has built this feature supplied by AI to display information supported by the best web results. Thus, it is neither “hallucinous” results like other AI programs.
AI’s overall problems come from other reasons, namely: misinterpreting queries, misunderstood online languages or lack information.
Google claims that these are challenges present in other research features. In addition, he has demystified some of the scandalous errors attributed to the new research function.
Certain viral screenshots say that AI previews have recommended leaving dogs in cars and smoking during pregnancy. However, Google says he tried the same requests and did not receive these ridiculous answers.
However, the company admitted that these errors can occur due to a “vacuum” or “information gap”. The latter is a lack of high quality content linked to a subject.
Google says that it does not “correct” the requests one by one. Instead, it updates the online research system to improve various types of request. Here are the most recent he has implemented:
- Improvement of detection mechanisms for absurd requests
- Limit the content generated by users for responses that can offer misleading advice
- Adding restrictions for requests where IA previews were not useful
- Improved guarantees for answers related to news, health and other important questions