{"id":252745,"date":"2021-05-18T20:58:03","date_gmt":"2021-05-18T17:58:03","guid":{"rendered":"https:\/\/en.buradabiliyorum.com\/googles-lamda-tech-will-bring-casual-conversations-to-google-assistant-review-geek\/"},"modified":"2021-05-18T20:58:03","modified_gmt":"2021-05-18T17:58:03","slug":"googles-lamda-tech-will-bring-casual-conversations-to-google-assistant-review-geek","status":"publish","type":"post","link":"https:\/\/buradabiliyorum.com\/en\/googles-lamda-tech-will-bring-casual-conversations-to-google-assistant-review-geek\/","title":{"rendered":"#Google\u2019s LaMDA Tech Will Bring Casual Conversations to Google Assistant \u2013 Review Geek"},"content":{"rendered":"

“#Google\u2019s LaMDA Tech Will Bring Casual Conversations to Google Assistant \u2013 Review Geek”<\/strong><\/p>\n

\n
\"The
Google<\/a><\/span><\/figcaption><\/figure>\n

Google Assistant has enough trouble understanding simple commands, but a breakthrough technology<\/a> called LaMDA<\/a> could help the AI have casual, free-flowing conversations. The LaMDA AI is still in development, but Google plans to open it for 3rd party testing in the near future.<\/p>\n

Demonstrated at the I\/O 2021 keynote, LaMDA (or Language Model for Dialogue App<\/a>lications), Google built LaMDA on its Transformer<\/a> open-source neural network architecture. Transformer is usually used for practical applications, like predictive text software and voice assistants. But Google trained LaMDA on casual dialog, giving it the ability to have free-flowing, human-like conversations on any subject.<\/p>\n

For whatever reason, Google chose to debut the LaMDA AI in two strange roleplaying scenarios. The first showed LaMDA speaking in the voice of Pluto, a self-described \u201cplanet\u201d that discussed current events like the New Horizons<\/a> mission, the coldness emptiness of space, and claims from scientists that Pluto isn\u2019t a real planet.<\/p>\n

The second conversation showed LaMDA in the role of a paper airplane. It described its job as a paper airplane and contemplated how the word \u201cgood\u201d could have many meanings\u2014a very zen response to the question of \u201cwhat makes a perfect paper airplane.\u201d<\/p>\n

These conversations were not programmed by Google, according to Google CEO Sundar Pichai. Instead, they were triggered by commands like \u201chello, I\u2019d like to have a conversation with a paper airplane.\u201d From there, the AI searches for information related to its role as an airplane, plus any topical paper airplane news<\/a> or adjacent topics (a paper airplane can get stuck in trees, so the AI might look up info on trees).<\/p>\n

Sundar Pichai mentioned that he and his son learned a lot talking to \u201cPluto.\u201d It\u2019s possible that the LaMDA AI may find early applications in education, as it will surely appeal to kids. A child who wants to learn about lions, for example, could ask LaMDA to pretend to be a lion. Either way, LaMDA is sure to have a big impact on conversational AI and Google Assistant.<\/p>\n

We will update this article with videos of Google\u2019s LaMDA demonstration when they\u2019re available.<\/p>\n

Source: Google I\/O 2021<\/a>, Google LaMDA Page<\/a><\/small>\n<\/div>\n