What do you want to know
- During Google I/O 2022, we saw a live demo of the AI Test Kitchen app for Android.
- The app offers three LaMDA 2 demos for in-depth conversations with Google Assistant: List It, Imagine It, and Talk About It.
- AI Test Kitchen is currently in closed beta, with Google “gradually” letting in new users.
- This crowdsourced testing app is designed to reduce “inaccurate or offensive answers” so the AI can stay “on topic”.
After unveiling its futuristic Language Model for Dialog Applications (LaMDA) at last year’s I/O, Google announced the new and improved LaMDA 2 at Google’s I/O 2022. And unlike last year’s on-stage demo, you’ll be able to test Google’s revamped AI and machine learning tools for yourself — eventually.
LaMDA’s goal is to let you have extended conversations with Google Assistant, where the AI stays on a particular topic or branches off to other topics based on your interest – simulating a proper conversation. It’s not (yet) available in the Google Assistant app, but you can test LaMDA in the AI Test Kitchen app that’s currently in closed beta.
Google’s goal with AI Test Kitchen is essentially crowdsourcing, which makes its AI more useful to people’s needs. He first sent the app to Google employees for feedback, which resulted in a “reduction of inaccurate or offensive responses” from LaMDA. Now, he will slowly open the app to everyday people to get their feedback as well and “responsibly learn, improve and innovate together on AI.”
In the app you will find three demos. list it gets you covered on a topic like “Planting a Vegetable Garden” and given a to-do list of what you’ll need to do or learn before you start; you can then tap on specific items to learn more or ask it to refresh the list with new ideas.
The second demo is imagine itand it follows the pattern of demo from last year: you ask LaMDA to describe an experience like visiting Pluto or the ocean floor, then you prompt it to take the story in different directions like “what is the temperature like” or “describe the jellyfish” in based on your interests.
Lately, Talk about it (dog edition) lets you chat about any dog-related topic you like. It’s not just to be cute; Google needs to test if the AI can remember the original discussion topic or if you can accidentally (or intentionally) trick it into forgetting how new questions relate to old questions.
You can label LaMDA responses as Nice, Offensive, Off Topic, or Not True, which should help the model learn to better anticipate our needs.
The difference between LaMDA and a smart speaker saying “Here’s something we found on the web” is that it’s designed to say things confidently and in a conversational way, so users expect statements to be true. This is why Google itself has stated that “there are significant challenges to resolve before these models can really be useful”, and why it will remain in beta for some time.
“While we’ve improved security, the model can still generate inaccurate, inappropriate, or offensive responses. That’s why we invite in-app feedback, so people can help report issues,” Google said. .
Hopefully it goes better than when Twitter users taught Microsoft’s chatbot to be racist in 2016, and users aren’t skewing LaMDA’s sense of “relevance” in the wrong direction. But assuming Is Alright, this will evolve Google Assistant beyond a question-and-answer tool into something much more useful.
Google Pixel 6
The best AI Google has to offer
Thanks to Pixel 6’s Google Tensor chip, Google Assistant’s responses are faster than ever, with tricks like Live Translate and speech-to-text transcription. And when LaMDA becomes part of the Assistant, you’ll get better search results than ever.
#Googles #Test #Kitchen #lets #teach #conversational #skills