Open in app
Sign inGet started
Model Alignment Through Automatic Prompt Updates From User Feedback

Model Alignment Through Automatic Prompt Updates From User Feedback

by James Wexler (PAIR Software Engineer) and Colby Hawker (Vertex AI Product Manager)
Go to the profile of James Wexler
James Wexler
Latest
Introducing the Conversational Toolkit

Introducing the Conversational Toolkit

by Adam Connors
Go to the profile of People + AI Research @ Google
People + AI Research @ Google
“Teach my son History” — How I used AI to help my son remember stuff

“Teach my son History” — How I used AI to help my son remember stuff

A week before his final exams and this is what my son’s bedroom floor looks like:
Go to the profile of People + AI Research @ Google
People + AI Research @ Google
Building the plane while flying it: an LLM case study

Building the plane while flying it: an LLM case study

Quinn Madison, Alexa Koenings, Iris Chu, Pratheek I, from Google Dev AI UX  Contributor: Laurie Pham, Google DevAI UX
Go to the profile of People + AI Research @ Google
People + AI Research @ Google
Talking to My AI (Part #2) — Is it useful yet?

Talking to My AI (Part #2) — Is it useful yet?

By Adam Connors
Go to the profile of People + AI Research @ Google
People + AI Research @ Google
Talking to My AI (Part #1) — Exploring the surprisingly useful illusion of human-like conversation

Talking to My AI (Part #1) — Exploring the surprisingly useful illusion of human-like conversation

The AI-Augmented life project is exploring ways that AI might change our everyday lives in the coming years. One theme that stands out in…
Go to the profile of People + AI Research @ Google
People + AI Research @ Google
Trending
Go to the profile of People + AI Research @ Google
People + AI Research @ Google

A snapshot of AI-powered reminiscing in Google Photos

Thryn Shapira discusses balancing user control…

Read more…
5 responses
Go to the profile of People + AI Research @ Google
People + AI Research @ Google
A smiling human form holding a green sphere in one hand stands next to drawer that has been opened from a wall of drawers. The greens sphere indicates a mental model. A burst of light-color and abstract shapes indicate that the sphere has been taken from the drawer. Two other random drawers are partially open, with green spheres peeking out.

Generative AI is reshaping our mental models of how products work. Product teams must adjust.

Read more…
2 responses
Go to the profile of People + AI Research @ Google
People + AI Research @ Google
LLM Comparator’s interface features a table on the left displaying input prompts, model outputs, score distributions, and rationales. A column on the right showcases visualizations like score distributions, metrics, clusters, and custom functions. Users can interact with both the table and visualizations to drill down into specific data points and filter results.

LLM Comparator: A tool for human-driven LLM evaluation

Read more…
1 response
Go to the profile of Slava Polonski, PhD
Slava Polonski, PhD

When to book and when to fly? Explaining prices in Google Flights

Read more…
1 response
Go to the profile of People + AI Research @ Google
People + AI Research @ Google
A GIF quickly showing off AutoNotes features. This begins on the “Chat” page. The user types in a query and clicks on the link to referenced note included in Gemini’s response. Next, the user clicks on the “Home” page and browses (via scrolling and uncollapsing containers) the hierarchical tags and summaries. Finally, the user clicks on “Highlights” for a fun preview of quotes, favorite people, etc. across the entire set of notes.

Adventures with AI-powered notetaking

Introducing AutoNotes: hierarchical tags, chat with your notes…

Read more…
1 response
About People + AI ResearchLatest StoriesArchiveAbout MediumTermsPrivacyTeams