Dobb-E Revolutionizes Home Robotics

PLUS: OpenAI explores method for controlling Superintelligent AI using imperfect instructions.

Welcome back. Today is Monday, December 18th and today we're looking at ways to ethically align generative AI, a system rapidly teaching robots household tasks, and big tech's ongoing investments in AI startups. New to The Intelligence Age? Sign up here.
Drop us a line anytime at [email protected] with feedback.

News & Insights

This New System Can Teach A Robot A Simple Household Task Within 20 Minutes

In a leap for domestic robotics, the Dobb-E system is now teaching robots household chores in just 20 minutes, overcoming the training data scarcity that hampers robots' learning. Unlike their digital counterparts, robots can't rely on the internet's inexhaustible data banks and require real-world interaction to gather information. Dobb-E's strategy, employing an iPhone on a stick, ingeniously amasses necessary data from actual home environments, enabling robots to perform tasks with an impressive 81% success rate across several New York homes.

The Dobb-E project, by integrating techniques like self-supervised learning, has allowed robots to quickly adapt to the varied and unpredictable nature of home settings. After a brief initial demonstration, these robots are fine-tuned to perform actions from pouring drinks to organizing recyclables. Challenges remain, such as dealing with reflective surfaces or heavy lifting, but the project's open-source ethos invites the global community to refine and expand this groundbreaking work. technologyreview

Eye On AI: Big Tech Continues To Invest In AI Startups, But For How Long?

The AI landscape remains dynamic as Mistral AI's staggering $487 million funding, led by Andreessen Horowitz with notable contributions from Nvidia and Salesforce, makes headlines. Nvidia's aggressive investment strategy, with 18 deals this year, underscores its commitment to shaping the AI ecosystem. Salesforce Ventures is not far behind, with a strategic focus on AI startups like Anthropics and Cohere. Meanwhile, Microsoft and Google continue to fortify their AI portfolios, the former with its monumental OpenAI deal and the latter with steady investments including a significant round for Hugging Face.

Regulatory scrutiny looms as the FTC examines Microsoft's OpenAI investment for potential antitrust implications. This scrutiny highlights the delicate balance between fostering innovation and maintaining competitive markets. Despite this, the trend of big tech nurturing AI startups persists, with the sector's need for substantial resources aligning perfectly with what these tech giants can offer. crunchbase

Together with Taskade

A second brain for you and your teams

Taskade is trusted by teams at companies including Lyft, TransferWise, Netflix, and Sony.

Get five AI-powered tools in one to help supercharge team productivity. 

  • AI Generator & Chat (replaces ChatGPT, Jasper, & CopyAI)

  • Tasks & Projects (replaces Trello, Asana, & Monday)

  • Notes & Docs (replaces Notion, Google Docs, & Obsidian)

  • Mind Maps (replaces Miro, Lucidchart, & Whimsical)

Try it now → (use code “intelligenceage” for 20% off)

Please support our sponsors!

Mastering AI

Top Tools

  • MyReport by alaba a.i.: AI-powered tool that automates the data collection and citation process for reports. Users simply describe the topic of their report, review it, and send. (website) (sponsored)

  • Peech: Automates video transcribing, editing, repurposing & branding for marketers. (website)

  • Photosonic AI Art Generator: A tool that generates unique images using AI technology. (website)

Practical AI

Around The World

More News

  • Microsoft's AI Chatbot Controversially Spreads Election Misinformation Including Conspiracy Theories and False Scandals When Responding to User Inquiries.

  • Former Google CEO Eric Schmidt Proposes a Comprehensive 6-Point Strategy to Combat Election Misinformation and Strengthen Democratic Processes.

Trending Today

Research Review

OpenAI Demos A Control Method For Superintelligent AI

OpenAI recently showcased an innovative approach to future AI control mechanisms, executing an experiment whereby the less capable GPT-2 supervised the more advanced GPT-4. This exercise aimed to explore how future superintelligent systems might be managed. While GPT-2 transferred its training to GPT-4, the latter, even with basic knowledge, surpassed its predecessor, particularly shining in natural language processing tasks. Such results suggest that a more powerful AI could, in theory, discern correct actions even when guided by imperfect instructions.

The significance of this research lies in its potential implications for the alignment problem—a key concern as AI capabilities outpace human intelligence. OpenAI's experiment indicates a possibility that superintelligent AI could autonomously align with human goals, even when receiving flawed guidance. This "weak-to-strong generalization" hints at a future where AI might independently avoid harmful actions, navigating complex tasks that exceed human understanding. siliconrepublic

Written by Isaac R. Ward, Casey Clifton, and Alex Brogan.

Send us feedback at [email protected] and help us provide the best coverage of Artificial Intelligence possible.

Interested in reaching smart business leaders like you? To become an Intelligence Age partner, apply here.

Join the conversation

or to participate.