Leadership CONNECT: 14-SEP-2024 (Y24W38)

Dear Friend,

This week’s Leadership CONNECT:

  1. New Tech: Microsoft LLMs Next Level: 1 Million Token Prompt (Source: Microsoft Research)
  2. Study: The Adoption of ChatGPT (Source: UChicago economists)
  3. Guide: 3 ways to spark AI adoption at work (Source: Work lab, Microsoft)
  4. Talk: The Creator Economy by Paul Saffo (Source: Long New Foundation)
  5. Thinking Update: Bulk Add Box (Source: MindLi)

Happy Thinking,

Dr. Yesha Sivan and the MindLi Team
P.S. Comments, ideas, feedback? Send me an e-mail

DigitalRosh > Knowledge > Technologies > AI-AGI

1. New Technology: Microsoft LLMs Next Level: 1 Million Token Prompt

(Source: Microsoft Research)

MInference is an AI tool designed to speed up and improve the performance of long-context language models (LLMs), which are used in applications that handle large amounts of text or data.

It leverages patterns in how LLMs focus on different parts of data (called “attention”) to reduce the time it takes to process large inputs, without compromising accuracy. MInference can achieve up to a 10x speedup in data processing using a single A100 GPU, making it highly efficient for handling million-token-level prompts.

Key Benefits:

  • Speed: Significantly reduces the time needed for LLMs to process large inputs, improving overall system efficiency.

  • Cost-Efficiency: Enables high-performance AI processing with lower infrastructure costs by using a single GPU.

  • Scalability: Effective for applications needing to process large volumes of data or long texts.

  • Accuracy: Maintains or slightly improves the language models’ ability to understand and retrieve relevant information.

Main Use Cases:

  • Enhances LLM performance in areas such as search queries, data retrieval, summarization, and coding tasks.

Main Risks:

  • Complexity: Requires advanced AI infrastructure and expertise to implement effectively.

  • Limited to Specific Hardware: Optimized for A100 GPUs, potentially limiting its use with other systems.

MInference makes LLMs more practical and cost-effective for businesses that rely on processing large datasets or lengthy text inputs, such as those in advertising, coding, or knowledge retrieval.

Read the full article on DigitalRosh 

DigitalRosh > Knowledge > Cross Industries > Future of Work

2. Study: The Adoption of ChatGPT

(Source: UChicago economists

Recent research from UChicago economists indicates that the AI chatbot ChatGPT is effective in various investment-related tasks and is expected to disrupt numerous high-skilled professions.

The study examines the adoption of ChatGPT among 100,000 workers across 11 occupations, revealing that half have used the tool, with usage rates varying significantly by occupation.

A notable gender gap exists, with women being less likely to adopt ChatGPT than men. While many workers recognize the productivity potential of ChatGPT, barriers to adoption remain, particularly for women.

The research combines survey data with administrative labor market information to analyze these trends.

Review the article on DigitalRosh

DigitalRosh > Knowledge > Cross Industries > Future of Work

3.  Guide: 3 ways to spark AI adoption at work

(Source: Work lab, Microsoft)

From a managerial perspective, AI adoption at Microsoft underscores the importance of achieving widespread organizational engagement to drive true business transformation.

Early adopters alone are insufficient; it requires fostering new habits across large groups through strategic initiatives.

Key strategies for accelerating AI adoption include:

  1. Employee-Driven Idea Sharing: Programs like “What a Great Idea” encourage employees to share insights on AI’s potential, resulting in a growing number of AI use cases and practical benefits that inspire further exploration.

  2. Gamification and Engagement: By making AI adoption enjoyable through gamified activities (e.g., scavenger hunts and challenges), the sales team fostered daily Copilot usage, which indicated deeper behavioral shifts.

  3. Skill-Building Initiatives: Organized learning events, such as skills-building series and dedicated learning weeks, create space for employees to experiment and develop AI competencies, resulting in notable increases in AI tool usage.

  4. Localized Community Events: Spark Weeks and face-to-face gatherings encourage team collaboration and context-specific AI learning, fostering both collective buy-in and cultural alignment.

  5. Cultural Integration of Learning: Embedding continuous learning and innovation as organizational values ensures AI adoption is sustained over time, turning curiosity into daily application.

In summary, sustained AI adoption requires not just technology deployment but also fostering a culture that values experimentation, shared learning, and practical engagement.

To review the full report on DigitalRosh 

DigitalRosh > Knowledge > Cross Industries > Innovation

4. Talk: The Creator Economy by Paul Saffo

(Source: Long New Foundation )

According to futurist (and Long Now board member) Paul Saffo, the “new economy” anticipated in the late 1990s is arriving late and in utterly unexpected ways.

Social media, maker culture, the proliferation of sensors, and even the 2008 market crash are merely local phenomena in a much larger shift.

What unfolds in the next few years will determine the shape of the global economy for the next half-century and will force a profound rethink of economic theory.

Paul Saffo teaches forecasting at Stanford and Singularity University. Journalists rely on him for cruelly telling quotes about everything from the monthly disruptions in Silicon Valley to the yearly turmoils in the global economy.

Watch the lecture on DigitalRosh

5. Thinking Update: Bulk Add Boxes
 

(Source: MindLi)

This section of the newsletter will provide you with a better understanding of the qualities of MindLi, share tips and tricks, and update you with new features, as well as other MindLi news. More: Help page or MindLi’s YouTube channel.

If you have a full project on your mind Bulk Add Boxes is the feature for you. 

You can create a full “tree” of Boxes and Nested Boxes in one place, using  + for Root Boxes, and ++. +++ for Nested Boxes. You will find this feature at the “More button” in the Plan/ Box view screen.     

This feature is currently available on our WEB application and on iOS. Soon in Android too. 

You are invited to test the new capability and share with us your thoughts & ideas.

How to Keep in Touch

Wish to Stay Updated? 

For updates on DigitalRosh activities and news, sent by the group manager only. 
Join our WhatsApp group

Share Your Thoughts With Us

We will be happy to hear from you with any feedback, ideas, or questions.

Contact us.

Join Our LinkedIn Groups

About Leadership CONNECT

MindLi newsletter, Leadership CONNECT, is a weekly source of news and inspiration for thinking aimed at DigitalRosh members and MindLi users. 

In every issue, you will find curated content from different industries. It would take less than 5 minutes reading to get a brief overview of the presented topics.

Want to delve deeper? You would need to have a subscription to the DigitalRosh site which is an affiliate site of MindLi. It is Free. Click on the links in each item or follow the breadcrumbs to the DigitalRosh website.

Enjoy!

See past CONNECT editions

Copyright © 2024 MindLi, All rights reserved.