What I learned at the GenAI Summit, May 2024

Microsoft for Startups Philip Rathle Soumya Batra Andreas Ramos Matthew Schwab Anil Katti Russell Miller Sharadon Smith Cedric Vidal Kay Gallagher

There are about 70,000 Artificial Intelligence companies worldwide. Many were represented at the GenAI Summit at the Palace of Fine Arts in San Francisco last week (May 29-31).

Microsoft, Amazon, IBM, and several others were the major conference sponsors. Google and Open AI (GPT-4 and, most recently, GPT-4o) did not seem to be actively present. GPT-4, however, is the foundation for some of the other large language models (LLMs) and tools.

I attended several activities during two of the three days. ?Chaos, massive crowds, serious sound, and technical problems in open areas made it hard to hear speakers.?So, this is what I got to hear and see. However, the Summit inspired me to dig deeper into my work on building prompts and technologies for marketing, sales, and business workflows.

What is Generative AI or GenAI?

Generative AI is a type of artificial intelligence technology that broadly describes?machine learning systems capable of generating text, images, code, videos, designs, analytics, or other types of content, often in response to a user's prompt.

News on the models, tools, and growth paths?

Many talks in keynotes, panels, and work groups covered all aspects of Generative AI, including the history, technologies, large language models (LLMs), Data and Data Testing procedures, applications, social aspects, and much more.

Soumya Batra, Philip Rathle, and Cedric Vidal were excellent speakers, but I missed other talks due to schedule changes or an inability to hear anything.

Soumya Batra, co-author of Llama3 and Llama 2 and with 11 years of development experience in Natural Language Processing, shared "LLMs: Present, Past, and Future." She articulately shared the history of LLMs from neural networks onward and her personal insights.

Philip Rathle, Neo4's CTO, was an excellent and inspirational speaker on the "AI and the Future of Work" panel and the host of the "Synthetic Data in the Era of AI" panel discussion.

Cedric Vidal, Principal Advocate for Startups @Microsoft, gave the keynote: "Improving RAG with Retrieval Augmented Fine-Tuning (RAFT) with AI Studio: Practical Insights," broke down leading-edge concepts, frameworks, and processes. RAG is?a technique that combines the capabilities of pre-trained large language models with external data sources, allowing for more nuanced and accurate AI responses.? RAFT is "Retrieval-Augmented Fine-Tuning," combining the benefits of Retrieval-Augmented Generation and Fine-Tuning for better domain adaptation. RAFT serves as the conduit between?generalized AI and domain-specific proficiency; domain-specific proficiency is a crucial concept.

Could you show me the money?

Hundreds of companies, schools, and exhibitors found customers and investors to expand. They use Gen AI to build domain-specific models and tools for Enterprises (EBs) and Small and medium Businesses (SMBs). Summit showed how GenAI can go way beyond the general public's use of GenAI using LLMs for all of the Internet.?Domain-specific models would have more control and a deeper context for answering questions. If AI companies can create sustainable business models, it promises to be a gigantic and massively lucrative opportunity.

But this also reminds me of the dot-com boom. In the 1990s, internet companies like Amazon, eBay, and Yahoo grew rapidly and excited investors. It led to an influx of venture capital, raising stock prices at first. The stock market made some people very rich but also led to bankruptcies and job losses. Alan Greenspan, the chairman of the Federal Reserve Board in Washington, then used the term "irrational exuberance" to describe the behavior of stock market investors. After the "boom," there was a sharp decline in early 2000.

What were the newer or better ideas for GenAI applications?

We consumers all experience various implementations of GenAI, such as conversational Chat GPT, AI Chatbots, prompting, content corrections, and suggestions. We also see ways to create images and do business analytics. The demos and workshops showed ongoing GenAI advancements.

Newer for me was taking static images and creating some movement in a video (Vimmerse); video production is still developing but moving fast. I also discovered the world of multiple AI agents.

Say one agent does a particular task, like autonomous agents in finance or software programs that trade, manage risk, and detect fraud in financial markets. Or in transportation: software programs that can drive cars and optimize routes. Healthcare: diagnose and treat patients. Multiple agents that automatically understand their environments make decisions and take action can hand off information to other AI agents.

What are the risks of using GenAI?

?I did not attend (or hear) a session on this topic. But I did see warnings and risks mentioned. ?The audience asked about risks, but speakers often gave general answers. There was no deep discussion about "risks" or security, copyright infringement, false results, unethical practices, or controlling what GenAI can access. A few panel speakers mentioned there would be a reduction or shift in the workforce for jobs that AI could take over. ??The Summit focused on tech advances, selling ideas, funding, shmoozing with technical leaders, following AI enthusiasts, and finding valuable contacts for more opportunities.

What were the fun activities?

In the circus spirit, we had a wonderful Robotic dog who begs (he won the crowds over as a dog!), AI toys for children, a robot that makes freshly filtered coffee, art shows with faces transitioning and spectacular and imaginative settings, There were fun Workshops to try out coding, too.

The best part was seeing real people and talking to them. Coffee was great, drinks were free for the last few hours, and conversations flowed freely, especially after the previous few hours of complimentary wine and beer.

"Ai is a tool to scale the imagination," Taryn Southern

What was the final outcome? I grew my understanding of Gen AI's fantastic potential. I expanded my practical and technical knowledge base. And I see the business models (subscriptions for consumers, major VC money, investments, and products for businesses to serve practical uses.)

We still need knowledgeable and ethical human experts who can sort the good, the bad, and the ugly, and make good decisions. GenAI is a great asset and tool, but we still need intelligent and wise people at the helm, having thoughtful AI conversations, automating processes, and eliminating malicious intent. We and the entire industry must know how to best protect our data and deal with the risks.

Did you attend the Summit? Please share your thoughts on the Summit and on GenAI in-general.

Note: This article was created by Marianne, corrected with Grammarly, and the image was a collage of photos with the help of Microsoft CoPilot and Designer. One big shout out to wonderful Professor Andreas Ramos, who gave me a ticket and shared his wit, wisdom, and understanding with many people at the Summit!