Since the beginning of this year, Google has been investing heavily in Artificial Intelligence (AI). From launching its AI chatbot, Bard, in February to giving Google search new AI-powered features, the tech giant has been all about the emerging technology. And now, Google has unveiled a new AI update for gmail, which is an attempt to fight spam.
The update is surely a much-needed one as spams are a concern for many gmail users and we’ve all been in situations wherein we ran out of storage due to those spam emails.
Google has made significant investments in Artificial Intelligence (AI) throughout this year. In February, they introduced their AI chatbot, Bard, and have since incorporated AI-powered features into Google search. Continuing their focus on this cutting-edge technology, Google has recently announced an AI update for Gmail, aimed at combating spam.
This update is highly anticipated, as spam emails have been a persistent issue for many Gmail users. We have all experienced situations where our storage was consumed by these unwanted messages.
RETVec, an advanced text classification system, forms the foundation of this improvement. According to Google, RETVec enhances the reliability and efficiency of text classifiers. It achieves exceptional classification performance while significantly reducing the computational resources required.
Text classification models play a vital role in popular Google services such as Gmail, YouTube, and Google Play. These models are crucial in detecting harmful content, including phishing attacks, inappropriate comments, and scams. However, they face challenges in classifying certain texts due to the tactics employed by malicious actors, such as homoglyphs, invisible characters, and keyword stuffing, to avoid detection.
One remarkable aspect of RETVec is its innovative architecture, which allows it to seamlessly handle all languages and characters without the need for extensive text preprocessing. This versatility makes RETVec suitable for various applications, ranging from on-device deployments to large-scale text classification on the web.
Google emphasizes that models trained with RETVec not only offer improved accuracy but also demonstrate faster inference speed due to their compact representation. Smaller models contribute to reduced computational costs and decreased latency, which are crucial factors for large-scale applications and on-device models.
Furthermore, Google has made RETVec an open-source tool, enabling developers to harness its capabilities in building resilient and efficient text classifiers for both server-side and on-device applications. Google specifically mentions that the Gmail spam filter already utilizes RETVec to strengthen its defenses against malicious emails.