Chatbots have been evolving since their inception in 1966. From ELIXA to ALEXA, AI-enabled conversation technologies have provided countless benefits to humankind. Google’s LaMDA foray into the industry six years ago might be the most groundbreaking yet.
Google has been experimenting with language for quite some time. More, Recently, inventing machine learning techniques that can grasp the intent of google searches. Since then, the company has made significant investments in the Research and development of Machine learning. LaMDA, their more recent project aims to solve the problems faced with intelligent artificial conversations.
Conversations lend to often wander from one topic to another in the blink of an eye. For example, a conversation about a movie can evolve into a discussion about the country where the movie was filmed. Such complex speech patterns are a predicament to modern conversation assistants, which follow a narrow, predefined path. But “Language Model for Dialogue Applications” or LaMDA can engage in a free-flowing conversation about a plethora of topics.
LaMDA Use cases:
LaMDA, like many other recent language models, is built on Transformer,(A neural network architecture of language understanding) invented by Google in 2017. Unlike other models, LaMDA bases its model on a metric called sensibleness. Sensibleness, like the word, suggests, checks if the response to a given conversational context makes sense. For example, if someone says “I just started going to the gym” you might expect a response like “that’s great, I remember when I first started working out, it was difficult but worth it”. This response makes sense based on the initial statement because rather than giving a generic response like “that’s nice” or “good for you”, the above response is context-specific.
Google has pledged to follow AI Ethics for their projects. Like many other technologies, AI can be misused. A language assistant can accidentally mirror hate speech or bias. Google has claimed they will try to minimize such risks and adhere to their AI Principles.
The Official description can be found at Google blog post