AI to Agent AI: Live from SudhaLive
Sudha Jamthe
Technology Futurist, Educator, GenAI Author, Researcher, LinkedIn Learning Instructor, Global South in AI, Stanford CSP & Business school of AI: IoT, Autonomous Vehicles, Generative AI
Thank You everyone who joined me for my monthly #SudhaLive Livestream today.
Today's topic was "Building Agent AI" and I hosted some students on Zoom and livestreamed on LinkedIn Live.
Agent AI now makes GenAI useful by allowing for Information Retrival (RAG) with agents offering ability to query any data to create useful, value creating workflow.
No more pretend GenAI creating copy and images with six fingers!
Here are some enterprise use case you can build with Agents from our session today!
As promised to all who joined #SudhaLive today, here is the session summary as generated by AI by Zoom.
Meeting Summary for Building Agent AI with Sudha Jamthe at #SudhaLive (from Zoom)
Quick recap
Sudha Jamthe discussed her approach to teaching AI courses and her interactions with students, introducing a new method of inviting students to join her on Zoom for a more interactive session. She also led a discussion on the concept of AI, focusing on building Agent AI, and highlighted the differences between traditional AI, generative AI, and agent AI. Sudha demonstrated the use of various AI tools, including AgentGPT, Akkio, Notebook LM, and CoPilot, and discussed the potential of AI in document and video analysis, emphasizing the importance of applying AI to create value in workflows.
Next steps
Summary
Teaching AI Courses and Building Agent AI
Sudha Jamthe, the host of a livestream on LinkedIn Learning, discussed her approach to teaching AI courses and her interactions with students. She introduced a new method of inviting students to join her on Zoom for a more interactive session and announced plans for new courses related to Agent AI on LinkedIn Learning and on Business School of AI. Sudha expressed her excitement about meeting new students and shared her experience of receiving comments from students who completed her course. She encouraged her students to join her Business School of AI community on LinkedIn, subscribe to her weekly newsletter, and attend her weekly webinars. Finally, she announced that the next part of the meeting would focus on building Agent AI.
Exploring AI Concepts and Limitations
Sudha led a discussion on the concept of AI, focusing on building Agent AI. She emphasized the importance of understanding AI practically, rather than just theoretically, and explained its two-step process of training with data and generating a software footprint. She highlighted AI's ability to recognize faces, voices, or predict weather patterns, but also its limitations, such as being narrow in understanding and prone to bias. Sudha also discussed the importance of AI's ability to recognize different voices, the need for proper training and testing of AI systems, and the limitations and potential failures of AI. She introduced the concepts of fairness and explainability in AI and encouraged the team to ask questions and further learn about AI.
Exploring AI Types and Agent AI Potential
Sudha discussed the differences between traditional AI, generative AI, and agent AI, emphasizing the shift from traditional AI to generative AI and the potential of agent AI. She also highlighted the popularity of ChatGPT and its underlying technology, GPT3.5. Sudha demonstrated how to build an agent AI using a no-code version, stressing the importance of choosing the right model and data sources. She also mentioned the potential for agent AI to be integrated into various platforms, such as Zoom, and offered to share more information on the topic.
Agent Capabilities and Decision-Making in Customer Interactions
Sudha discussed the role and capabilities of an agent, describing it as a Language Model with decision-making abilities. She explained that an agent can use reasoning to understand a customer's goal and interact with them, either autonomously or through iterative cycles. Sudha also mentioned that an agent can use tools and data stores to enhance its performance. She demonstrated the use of AgentGPT, a tool that can perform various tasks such as research, brand performance planning, and event planning. The tool breaks down the task into smaller tasks and uses an LLM to compile the information into a report format.
领英推荐
Data Sources, LLMs, and Storage Discussion
Sudha discussed the use of data sources for various tasks, including storage and LLM needs, and the importance of choosing the right LLM for specific tasks. She also addressed questions about storage, privacy, and security, emphasizing the need for trust in the chosen options and the importance of considering where data is stored. Sudha introduced the tool Akkio, which integrates GPT and allows users to query datasets and draw insights, and highlighted the concept of retrieval augmented generation (RAG), which allows data sources to be kept separate. She encouraged participants to ask questions and promised to answer them, either herself or through her team.
Exploring AI Tools and Privacy Features
Sudha discussed the use of experimental AI tools, including NotebookLM and Copilot, emphasizing their privacy features and the importance of understanding AI basics before using them. She demonstrated how to create a new notebook and add a data source, and encouraged participants to try these tools on their own. Sudha also showcased NotebookLM that can generate summaries and study guides from documents, demonstrating its effectiveness with a book and a research paper.
Exploring AI in Document and Video Analysis
Sudha discussed the potential of AI in document and video analysis, using examples such as Meta prompting and AI reasoning. (citation of paper showed: Meta Prompting for AI Systems by Yifan Zhang, Yang Yuan, Andrew Chi-Chih Yao doi.org/10.48550/arXiv.2311.11482)
She introduced 'AgentGPT' and 'Notebook LM', and highlighted the efficiency of AI in finding data sources and summarizing information. Sudha emphasized the importance of applying AI to create value in workflows such as sales, marketing, business analysis, and HR. She encouraged the team to experiment with these tools to generate summaries from various data sources, including videos and audio.
Sudha also expressed her enthusiasm for helping her students and invited everyone to join her weeklywed sessions online and promised to notify them when her AgentAI course is available on LinkedIn Learning.AI
== end Zoom AI generated summary ==
I enjoyed your questions and our #SudhaLive session today. Thank You everyone for joining, I appreciate our time learning together.
Post your use case or questions Business School of AI and I'll answer your questions if you need help.
Remember I am here to help!
Learn from Sudha Jamthe:
Sudha Jamthe is a Technology Futurist, educator and founder of Business School of AI. She is Top voice in Artificial Intelligence on LinkedIn and enjoys mentoring business leaders to innovate. She enjoys chasing self-driving cars and hugging robots.
? Research / Data Scientist ? Data Visualization Expert (BI) ? Data Engineer ? Machine Learning ? Data Analysis ? Deep Learning ? Cloud Technologies ? MLOps ? Python ? R ? SQL ? Power BI ? Tableau
5 个月Alex Frebergto
Honorary investment counsellor
5 个月Very informative