Google’s AI wants to be smart in any case

Google has designed an AI that is capable of interacting in a natural way – Gecko

Where is the LaMDA Project? Google took advantage of its annual I / O conference to unveil its progress in artificial intelligence.

LaMDA (Language Model for Dialogue Applications) is an AI capable of understanding interaction. The bot can pretend to be anything and everything, and can start discussing any topic in a natural way. The final idea is to program LaMDA chatbots (chatbots) that can pretend to be a business, person, entity, or whatever. To better explain the concept of this AI, Google took the example of a discussion with the planet Pluto, and then with a paper airplane.

LaMDA is able to improve reactions by pretending to be Pluto, this is very surprising. If AI can engage in any interaction without the need for predefined scripts, it is thanks to many concepts that will have stored and learned it. It keeps the dialogue open without going around in circles. In Pluto’s example, “no answer is predefined,” says Google boss Sundar Pichai. Dialogue seems to flow naturally.

In the fetus

This chatbot differs from what we already know with Alexa, Siri, or the current Google Assistant, as the goal is to use LaMDA verbally in fluid conversations: “Unlike other language models, LaMDA is trained on dialogues During his training, he noticed various nuances that differentiate conversation from other forms of language, “Google explains on his blog.

For now, AI is still in its infancy and has trained only on lessons. The next step will be to be able to simultaneously detect information via text, audio, video and image. This will allow questions to be asked through various sources of information, as summarized, in conversation on traditional messages.

In addition, Google has stated that it is attentive to the biases of such AI and tries to limit the risks. In fact, an AI is content to interpret the data given to it, so it can take into account biases, hate speech and misinformation, as has happened many times before. In particular, the Microsoft Twitter chatbot became deeply racist just 24 hours after entering the social network.

The Mountain View firm said that she was eventually tempted to add LaMDA to Google Assistant. In addition, the bot will certainly be intended for other services and businesses. Anyway, LaMDA is still only in the internal project phase at Google, and the firm has not yet provided information to the general public about the possible adaptation of AI software.

See also  Fortnite leak promises Season 4 "may well" have a comprehensive Marvel concept

You May Also Like

About the Author: Tad Fisher

Prone to fits of apathy. Music specialist. Extreme food enthusiast. Amateur problem solver.

Leave a Reply

Your email address will not be published. Required fields are marked *