AI Con USA 2024 - Concurrent Sessions
Concurrent sessions offer attendees the flexibility to explore a variety of topics throughout the conference on Wednesday and Thursday in order to customize their learning experience. Learn both enterprise foundations and new methodologies to grow your skills, supercharge your knowledge, and re-energize your career growth.
Wednesday, June 5
How NOT to Use GitHub Copilot: The Time Traveler's Playbook for Coding with AI
"Does GitHub Copilot stunt learning, and why doesn't it always produce relevant suggestions?" These concerns echo the anxiety surrounding using AI to code, often rooted in the steep learning curve associated with adopting new technology. What if you were told that you might be using GitHub Copilot wrong? Through 2000s nostalgia and interactive storytelling, Rizel will provide attendees with a practical framework for effectively using GitHub Copilot, so you can write code that elevates your career and learn new programming languages. Audience members will leave with strategies for using AI...
AI Risk and Safety: Proactive Monitoring for the Future of Humanity and Business
In an era where AI's capabilities are rapidly advancing, the question of safety and risk management becomes paramount. While AI currently serves as a beneficial tool, its potential evolution poses significant risks. The possibility of AI surpassing human intelligence, outperforming human capabilities, or developing biases misaligned with human values raises concerns about existential risks comparable to, or potentially exceeding, those of nuclear warfare. High-level attention, including inquiries from Congress and presidential directives, underscores the gravity of these concerns. Some...
Troubleshooting and Resolving Common LLM Issues In Production
According to a recent survey, 61.7% of enterprise engineering teams now have or are planning to have a generative AI application within a year—and 14.1% are already in production. As enterprises race to deploy generative AI into their businesses, the need to ensure that LLMs are deployed reliably and responsibly is paramount. But how can enterprises and AI engineers evaluate and troubleshoot models in real time? In this session, Amber Roberts, a data scientist and machine learning engineer at Arize AI, will cover emerging best practices from direct work advising enterprises with real...
The Rise of the Virtual QA Engineer: Harnessing GenAI for a Productivity Boost—A Production Case Study
In this talk, we will delve into the transformative journey of integrating GenAI into the core of our testing and development processes. This integration has not only enhanced our productivity by 15% but also yielded a 20% time savings and a significant cost reduction per test case. Want to know how? We'll explore the strategic implementation of GenAI, overcoming security challenges, leveraging diverse (LLMs), and the meticulous design of prompts that culminated in a prompt library. Our tailored extensions for the code editors and corporate chats exemplify the seamless fusion of AI with...
Production Level Prompting & Techniques for LLMs
Let’s dive into what goes into creating expert-level prompts for ChatGPT and other LLMs. We will focus on the key ideas to take into account when prompting for any use case and how to better understand the results you get back. We’ll look at prompting 101, think about language, and explore advanced prompting frameworks. Some of the challenges we will address include: how to improve your results, how to process large text, why ChatGPT doesn't do what you want, what your prompt should look like, and many more!
Postman and AI: Transforming Manual QA to Quality Engineers
Facing the dual challenge of enhancing deployment quality without extending regression testing time, Ethan Ferguson's team turned to automation. The team consisted mainly of manual QA professionals with a background in UI testing, not automation engineering. The question was how to leverage the existing team's skills to test services layers and develop stable release automation scripts. Utilizing prior experience in building Postman automation frameworks, they aimed to empower the team. Concurrent company changes included adopting Postman as the primary API testing tool and integrating...
Exploit Grandma: Prompt Better with Cloudflare AI
Want to get better output from Large Language Models (LLMs)? Give it better input! This talk will go over techniques to help you be a better prompt engineer.
Mad Data Science: Using AI to Build Ridiculous Things
PreviewHave you ever gotten into an argument on whether or not Die Hard is a Christmas movie and thought "If only machine learning could help" or "What could I do to automate my dog's daily routine and make his life easier" or "Is it possible to spend 3 months time to determine if I can sleep in 5 minutes tomorrow morning"? If so, you might get along with Matt Eland. Come join this lighthearted talk as he tours you through some ridiculous uses of AI and Machine Learning with topics ranging from large language models to reinforcement learning to simple regression algorithms. You'll get a...
Thursday, June 6
Fully Autonomous Software Testing with GenAI
For decades, the concept of software autonomously testing itself has been a tantalizing dream. The traditional approach to software development involves extensive human effort in clicking through applications, writing API and UI test scripts, and a multitude of manual tasks before releasing a new version. However, the rise of Generative AI (GenAI) is heralding a transformative era in software testing, enabling a future where machines can undertake the bulk of these testing tasks. GenAI's capabilities extend beyond creating music, images, and even product code—it now encompasses the...
Prototype at the Speed of Thought: How AI is Changing the Game for Creators
Most product ideas die on the backlog because we can't find the time, lack the skills, or fail to get others excited about them. But thanks to AI, that no longer has to be the case. Product Builders, Project Managers, DevOps Managers, and simply creators of all kinds can turn a napkin sketch into a clickable prototype within hours. Rolf Mistelbacher shares why AI is turbocharging prototyping and how the arrival of new AI-powered creator tools is driving that change. He shows how you can transform your idea into a clickable prototype without writing code—creating wireframes and a user...
AI Agents: Orchestrating the Future of Digital Workflows
The advent of AI Agents heralds a new era in computing, introducing a paradigm shift from conventional application-specific interactions to a holistic orchestration of digital workflows across various platforms and industries. This presentation will explore the concept of AI Agents as a pivotal innovation in artificial intelligence, capable of transforming user experiences by enabling seamless integration and management of tasks across multiple applications. Unlike their predecessors, AI Agents excel in providing a personalized, context-aware interface, learning from user interactions to...
Game Over—AI Steals Data, Is Biased, and Produces Deep Fakes
It has been two years, and ChatGPT and other Generative AI tools have taken the world by storm. We all still have jobs. Phew! However, AI is still wreaking havoc in our world. Join Kevin as he takes you on an adventure through bias, deep fakes, and privacy concerns. This rollercoaster of a presentation will be fun, scary, and hopefully…well…maybe you will have hope by the end—either way, you won’t want to miss this lively presentation full of controversy, intrigue, and mystery. Wait, is this Clue or a presentation? Guess you will have to join us to find out!
How Gen AI and Prompt Engineering Turns Testers into Superheroes!
Envision a testing realm infused with AI-driven superpowers, where a reliable AI companion will revolutionize how we approach testing. This companion doesn't just assist; it transforms the adventure of creating test cases that span the full spectrum of possibilities and pinpointing bugs in the most unexpected places. This is the essence of AI-augmented testing, a paradigm where AI elevates testers to maestros, orchestrating a symphony of enhanced augmented test intelligence. Jonathan Wright will go beyond merely discussing the technological underpinnings (e.g., GNN, RAG, RGA, NLU). It's a...
Tips for Building LLM Apps
The barrier to entry for AI, specifically Large Language Models (LLM), is low and we can take advantage of them in our own tools, projects, and apps. In this session, we’ll build a simple app that uses an LLM which will help us cover: The steps to using AI in a program. LLM Basics. Pick the right LLM. Data Preparation. Evals and Testing. Insights and issues to consider. Tips that you can use as you build your own. Join us for a hands-on session that will help you build your first LLM app or level up your current project!
Prompt Engineering for Mere Mortals
Prompt Engineering is a skill that anyone can develop, even mere mortals. While Generative AI can produce amazing results, it isn’t magic. You don’t need to be an AI Wizard to learn to craft effective prompts that help you leverage Generative AI in a way that helps you complete your tasks and achieve your goals and objectives. Tom will break down the terms and use everyday examples to illustrate Prompt Engineering best practices so that you can more easily navigate the nuances of Generative AI. In this session, you will learn what Prompt Engineering is and how to use it to get the best...
Keeping AI Models Relevant in the Presence of Drift
How does an evolving environment affect the predictions of a Machine Learning (ML) system? Often, the characteristics of real-world environments change as time passes. When we build an AI system that looks for behavioral markers for a certain condition, what do we do when the behavioral markers change? They are no longer a good predictor, and we need to retrain our predictor system. As an environment changes and evolves, its evolutionary path, sometimes called 'drift,' can affect how a system in that environment functions. Environmental drift may adversely affect how well an ML system...