Business Today
Loading...

Google IO: LaMDA is the perfect example of how Google is cracking natural conversations but it is just a start

Google showcased its breakthrough conversation technology LaMDA at the ongoing IO 2021. The technology promises to bring the next step of evolution to AI conversations across Google services.

twitter-logoSarthak Dogra | May 18, 2021 | Updated 23:52 IST

Highlights

  • Google is working on LaMDA, a new conversation technology that is not restricted to pre-fed answers.
  • LaMDAis essentially a machine learning techniquethat brings about a conversational flow to an interaction with a user.
  • LaMDA is still under research and is being tested internally by Google as of now.

Google IO 2021 has begun and Google has already announced a series of updates to most of its online services. Amid these updates, the company announced LaMDA, its breakthrough conversation technology, that will soon empower Google Search and other Google-owned platforms.

LaMDA is essentially a machine learning technique that enables the next step to chatbots. While chatbots are usually limited to some pre-fixed responses, LaMDA will bring about a conversational flow to an interaction with a user.

Short for "Language Model for Dialogue Applications" LaMDA can engage in a free-flowing conversation that might involve a series of related or unrelated topics. Google believes this new ability to think can unlock more "natural ways of interacting with technology and entirely new categories of helpful applications."

As per Google, LaMDA builds on earlier Google research published in 2020. At the time, the research showed that Transformer-based language models trained on dialogue could learn to talk about virtually anything. Since then, Google found that upon training, LaMDA can be fine-tuned to significantly improve the "sensibleness and specificity of its responses."

LaMDA is still under research and is being tested internally by Google as of now. An example of the capabilities of LaMDA, as shared by Google in a blog, goes like this -

User: Hi, I'd like to have a conversation demonstration for a blog post.

LaMDA: Hi. I'm a friendly and knowledgeable demonstration for your blog post. I can engage in interesting conversation about virtually anything. What would you like to chat about?

User: if my washing machine and tv are both malfunctioning at the same time what could the reason be?

LaMDA: It could be the power surge from the washing machine may have fried your tv. If it's a power surge problem it may be time to get a new washing machine. Or a new TV.

User: yeah maybe. btw how many legs does a horse have, and could it fix my washing machine?

LaMDA: Horses have four legs. How could they help?

User: you tell me

LaMDA: Oh, well they could certainly help you. By moving you to a new home where you could install a new washing machine.

As can be seen, LaMDA's responses make sense as per the initial query. Though these responses are not limited by the sensibleness but also tend to be specific, by relating clearly to the context of the conversation. In the example above, the responses are sensible and specific.

  • Print
  • COMMENT
BT-Story-Page-B.gif
A    A   A
close