Recent deposition testimony revealed that Google utilizes its search engine data to train its Gemini AI models. This process involves using search signals to promote high-quality, authoritative content while minimizing low-quality, spammy sources. Additionally, search data has contributed to the development of the AI Overviews feature in Google Search, with user feedback playing a crucial role in refining its performance. This confirms earlier statements made by Google about employing search algorithms to elevate the quality of AI interactions.
In a separate internal email relating to training Google's Gemini model, a Google employee wrote that search "signals will be very helpful for us to upweight good authoritative pages and downweight the spammy untrustable ones."
The lawyer, Karl Herman, also showed deposition testimony from Google senior director of engineering Phiroze Parakh, who said that search data was used to pretrain the model that generates the AI Overviews feature in Google Search.
Collection
[
|
...
]