ChatGPT: Possibilities and Pitfalls

Jonathan Hassell: Introduction (5 minutes) - 8:00am PT | 11:00am ET | 3:00pm UTC/GMT

• Jonathan Hassell welcomes you to ChatGPT: Possibilities and Pitfalls.

Jordi Ribas and Tim O’Reilly: Reshaping Search with Bing (45 minutes) - 8:05am PT | 11:05am ET | 3:05pm UTC/GMT

• The new AI-powered Bing, a product of Microsoft’s recent partnership with OpenAI, has disrupted the status quo in the field of search engines. Referred to by Microsoft’s consumer chief marketing officer Yusuf Mehdi as “an AI copilot for the web,” Bing’s AI-powered abilities are reshaping search by providing an interactive chat-like experience and enabling quick content creation, from text to images. Join Tim O’Reilly and Jordi Ribas, corporate vice president of search and AI at Microsoft, for a prerecorded video discussion of the technologies behind Bing and Microsoft’s approach to building a “responsible AI ecosystem” that is changing the way that we interact with the internet.
• Jordi Ribas is corporate vice president of search and AI at Microsoft, leading the product and engineering teams responsible for Bing search worldwide as well as Microsoft Search in Bing for enterprise. Since 2008, Jordi has led multiple product and engineering teams focused on AI products and services such as search, speech, vision, and digital assistants. He was on point for Bing product and growth in 2014 and took on all of Bing in 2020. In February of 2023, he and team released the new Bing, which was the first large-scale search engine with search-grounded GPT answers and multiturn chat-based search, launching a new era of AI-powered search. Previously, Jordi led a variety of product, marketing, and business development teams in the US and Japan that delivered digital media software for Windows Media, Xbox, and CE products; did research on data compression in the digital video department at Sharp Labs of America; and worked as a researcher at NTT, Japan. He holds a PhD in electrical and computer engineering from the University of Michigan, Ann Arbor.

Thomas Dohmke and Tim O’Reilly: The Future of Developer AI and Copilot X (30 minutes) - 8:50am PT | 11:50am ET | 3:50pm UTC/GMT

• Join GitHub CEO Thomas Dohmke and Tim O'Reilly for a sweeping conversation on the future of generative AI. The two will go in depth on the emerging technologies that are shaping the next great technological advancement of our times, from ChatGPT to GitHub Copilot to the newly announced GitHub Copilot X.
• Fascinated by software development since his childhood in Germany, Thomas Dohmke has built a career building tools developers love and accelerating innovations that are changing software development. Thomas is chief executive officer of GitHub, where he oversaw the launch of the world's first at-scale AI developer tool, GitHub Copilot—and now, GitHub Copilot X. Previously, Thomas cofounded HockeyApp and led the company as CEO through its acquisition by Microsoft in 2014. He holds a PhD in mechanical engineering from the University of Glasgow, UK.
• Break (5 minutes)

Malte Pietsch: Connecting GPT with Your Data—How to Tailor Responses to Your Use Case and Avoid Hallucination (Sponsored by deepset) (20 minutes) 9:25am PT | 12:25pm ET | 4:25pm UTC/GMT

• Learn how to build with LLMs, like ChatGPT, and avoid typical pitfalls, such as hallucination, outdated information, and lack of understanding of domain-specific data. Malte Pietsch demonstrates how to connect LLMs to your data using retrieval-augmented generation. You’ll discover how to design prompts that minimize hallucination and how to evaluate the performance of your NLP application as you explore practical code examples, development workflow best practices, and traps you may encounter along the way.
• Malte Pietsch is the cofounder and CTO at deepset, where he builds Haystack and deepset Cloud to enable developers all over the world to use NLP effectively in their business applications. Previously, he conducted NLP research at Carnegie Mellon University and was a data scientist for multiple startups. He's been crafting NLP applications for all kinds of businesses for more than eight years now and is convinced that the development workflow and user orientation are the key criteria for successful NLP projects.
• This session will be followed by a 20-minute Q&A in a breakout room. Stop by if you have more questions for Malte.

Matt Welsh: Building Applications Using Large Language Models (30 minutes) - 9:45am PT | 12:45pm ET | 4:45pm UTC/GMT

• The world is freaking out about ChatGPT and the seemingly endless possibilities of large language models. But did you know that with just a little bit of work, you can teach models like GPT-3 how to communicate with external systems, data sources, and APIs? You can even get ChatGPT to converse with multiple versions of itself, each tailored for a different task or problem domain. In fact, large language models seem to represent a new kind of computational engine, one that might replace conventional software. Matt Welsh offers an overview of how to build new applications with LLMs at the core, using the Fixie platform.
• Matt Welsh is the cofounder and CEO of Fixie.ai, a startup developing a new computing platform for AI-based applications. Previously, he was SVP of engineering at OctoML and spent time at Apple and Google and as a professor at Harvard.
• Break (5 minutes)

Blaise Agüera y Arcas: Reassessing Intelligence—Insights from Large Language Models and the Quest for General AI (30 minutes) 10:20am PT | 1:20pm ET | 5:20pm UTC/GMT

• Large language models have now achieved many of the long-standing goals of the quest for general AI, and they’ve done so with large-scale neural machine learning, as opposed to the code- and rule-based approaches that have repeatedly failed over the past half century. While LLMs are still very imperfect (though rapidly improving) in areas like factual grounding, planning, reasoning, safety, memory, and consistency, they do understand concepts, are capable of insight and originality, can problem-solve, and exhibit many faculties we have historically defended vigorously as exceptionally human, such as humor, creativity, and theory of mind. At this point, human responses to the emergence of AI seem to be telling us more about our own psychology, hopes, and fears than about AI itself. However, taking these new AI capacities seriously, and noticing that they all emerge purely from sequence modeling, should cause us to reassess what our own brains are doing, and whether we’re learning what intelligence—machine or biological—actually is. Join Blaise Agüera y Arcas to explore this area more deeply and discuss some of the implications.
• Blaise Agüera y Arcas is a VP and fellow at Google Research, where he leads an organization working on both basic research and new products in AI. His focus is on augmentative, privacy-first, and collectively beneficial applications, including on-device ML for Android phones, wearables, and the Internet of Things. One of the team’s technical contributions is federated learning, an approach to training neural networks in a distributed setting that avoids sharing user data. Blaise also founded the Artists and Machine Intelligence program and has been an active participant in cross-disciplinary dialogs about AI and ethics, fairness and bias, policy, and risk. Until 2014 he was a Distinguished Engineer at Microsoft. Outside the tech world, Blaise has worked on computational humanities projects including the digital reconstruction of Sergei Prokudin-Gorskii’s color photography at the Library of Congress and the use of computer vision techniques to shed new light on Gutenberg’s printing technology.

David Wu: Morgan Stanley Wealth Management: Our GPT Journey (30 minutes) 10:50am PT | 1:50pm ET | 5:50pm UTC/GMT

• Join David Wu to explore Morgan Stanley Wealth Management’s recent work with OpenAI to launch a first-to-market financial advisor tool that leverages GPT-4 and Morgan Stanley’s vast intellectual capital to deliver relevant content and insights into the hands of advisors in seconds, helping drive efficiency and scale.
• David Wu is a managing director at Morgan Stanley and WM head of knowledge management, where he’s responsible for improving the experience for WM employees in finding accurate and high-quality information with speed and precision. As part of this mandate, his team is responsible for the content curation and oversight processes, the content platforms including the “3DR” portal leveraged by WM, the content management system (Adobe), search engines (Lucidworks), and leading-edge technologies such as the virtual assistant (Kore.ai) and NLP (OpenAI). Previously, David was the COO for WM Analytics and Data, where he was responsible for business administration and group strategy; held analytics, business development, and data strategy roles at Credit Suisse; and provided business and technology strategy expertise at Deloitte. David graduated from the Stern School of Business at New York University with dual majors in finance and accounting.
• Break (5 minutes)

Adam Witwer, Lucky Gunasekara, and Andy Hsieh: Using LLMs to Enhance Personalized Learning (30 minutes) 11:25am PT | 2:25pm ET | 6:25pm UTC/GMT

• In 2020, O’Reilly and the Miso team partnered in the release of Answers, a question-and-answer semantic search engine that was cross-trained on BERT and with five million question-and-answer pairs posed by programmers online, before being tuned on the entire corpus of the O’Reilly book library. Join them as they explore applying the latest advancements in LLMs to possible updates not just to Answers, but across the oreilly.com reader experience. The team will unpack topics including reconciling LLMs with the Credit, Consent and Compensation Model of oreilly.com; avoiding LLM hallucinations and adversarial prompting; fact checking an LLM and developing reasoning heuristics; balancing private, expert domain knowledge with an LLM’s open source “knowledge”; responsibly marrying personalization and LLMs; the potential for assistive AIs in learning; and more.
• Adam Witwer leads the product team at O'Reilly and has worked at the company for over 17 years. He loves creating products and experiences that help people and companies level up their technical skills.
• Lucky Gunasekara is the cofounder and CEO of Miso.ai, where he leads the team's efforts to create new search and discovery experiences that dissolve the barriers between content and readers. With a background in interaction design and user experience research, Lucky collaborates with publishing partners and newsrooms to develop seamless content search and discovery interfaces that are in balance with their editorial and journalistic standards. A longtime supporter of privacy-first systems, his core focus is making sure the next generation of AI and LLM systems are responsibly wielded, with a strong emphasis on credit, consent, and compensation for experts, publishers, and creators.
• Andy Hsieh is an engineering leader and cofounder and CTO of Miso Technologies Inc., where he combines his extensive expertise in machine learning with a pragmatic engineering mindset and astute product sense to create search and personalization systems that elevate user experiences for renowned publishers and ecommerce groups. With a proven track record in democratizing big-tech search and personalization using LLM and smarter and safer models, Andy's pioneering work has been published in esteemed conferences and journals. He’s dedicated to building machine learning solutions that prioritize user privacy and safety while delivering amazing user experiences.

Jonathan Hassell: Closing Remarks (5 minutes) - 11:55am PT | 2:55pm ET | 6:55pm UTC/GMT

• Jonathan Hassell closes out today’s event.

Sponsored by: