kkeynews.aikeynews.ai

September 2020

Previous month·Next month
keynews.ai

Key AI news stories in enterprise and tech.

For practitioners and enthusiasts

TodayThis WeekThis Month

Edition · September 2020

k.Frontier
FrontierOpenAI Blog
KeyRank 72

Learning to summarize with human feedback

This foundational research on RLHF (reinforcement learning from human feedback) became the training methodology behind ChatGPT and modern LLMs. It's a capability breakthrough that shaped how the entire industry trains alignment into models.

Why it ranks · · RLHF training methodology applied to summarization tasks · September 2020

Read full story
The six pillarsRanked by KeyRank · this month

Frontier

44% share

Models, benchmarks, the lab race. · 4 stories

  1. 1Learning to summarize with human feedback72
  2. 2Learning to segment images without manually segmented training data
  3. 3How to reduce annotation when evaluating AI systems
All 4 stories

Tools

33% share

The AI products people actually use. · 3 stories

  1. 1New Alexa features: Natural turn-taking65
  2. 2New Alexa features: Speaking style adaptation
  3. 3New Alexa features: Interactive teaching by customers

Agents

quiet today

Autonomous systems in the wild. · 0 stories

  1. Nothing cleared the bar today.

Money

11% share

Funding, M&A, valuations, earnings. · 1 story

  1. 1OpenAI licenses GPT-3 technology to Microsoft65

Chips

quiet today

Silicon, racks, the compute buildout. · 0 stories

  1. Nothing cleared the bar today.

Work

11% share

Jobs, industries, people & policy. · 1 story

  1. 1Aravind Srinivasan: Amazon Scholar focuses on combinatorial optimization, algorithms, and AI45
Enterprise & data stackSalesforce, Databricks, Snowflake & more · this month
  1. 1
    Learning to segment images without manually segmented training dataSegment
    45

Get it by email.

Daily · Weekly · Monthly