Google I/O Updates: On Wednesday, Google announced the debut of LaMDA 2, a new conversational artificial intelligence (AI) model that can help make discussions via computing devices easier and more efficient, at its I/O 2022 consumer keynote. LaMDA 2 is the successor to the original LaMDA — short for Language Models for Dialog Applications — which was introduced at the I/O developer conference last year. Google also unveiled AI Test Kitchen, a specialized platform for testing AI models such as LaMDA with people outside of the firm, alongside LaMDA 2.
About Google I/O
LaMDA 2, like the previous model, is designed to aid in the development of better conversational apps.
At the speech, Google CEO Sundar Pichai stated that LaMDA 2 had “amazing conversational capabilities” that can help make computer discussions more natural. To communicate with people on numerous issues, the new AI model incorporates a large number of datasets from multiple sources.
With an onstage presentation, Pichai demonstrated the reach of LaMDA 2 by asking the AI model to describe a scenario in the Mariana Trench, which is located in the western Pacific Ocean. The model answered by describing the organisms that inhabit the trench. It even asked follow-up questions and broached themes like submarines and bioluminescence that it hadn’t been expressly taught to address.
Maybe you’re interested in learning more about how smell works. You’ll receive a different response for that as well “Pichai explained.
He admitted that, despite the improved safety of LaMDA 2, the model could still “produce erroneous, inappropriate, or offensive responses.” For all of these reasons, Google created the AI Test Kitchen app, which allows professionals such as AI researchers, social scientists, and human rights experts to contribute to ongoing AI advances such as LaMDA.
Google has also made it possible to submit comments to its developers via the AI Test Kitchen app, allowing them to improve the AI model.
The AI Test Kitchen will be limited to the United States at first, and not all users will be able to access it. Its primary goal is to improve the testing models that Google has developed. The app will also continue to include other “developing areas of AI” in order to improve existing and future models.
Google is also developing a technique called chain-of-thought prompting to assist AI models answer even difficult questions like how many hours are there in a month by describing multi-step tasks as a sequence of intermediate steps.
In the area of AI and machine learning advancements, Google is not alone; Amazon, Meta, and Microsoft are all making significant attempts to expand their businesses in these growing fields. Google’s new initiatives, on the other hand, are intended to set it apart from the competitors. The company also has a number of tools available to assist it acquire enormous datasets and persuade a large number of people to help it test ongoing advancements efficiently.