GPT Builder: The Biggest Thing Since ChatGPT

Welcome back to the AI Geekly, by Brodie Woods. We’re taking a beat this week to talk about a sea change in the consumer AI space: OpenAI’s newly introduced (and confusingly-named) “GPTs” — user-customizable flavors of its ChatGPT platform that opens-up bleeding edge AI capabilities to the masses in a no-code, user-friendly chatbot envelope. It’s hard to overstate how transformational this will be. This is AI’s iPhone moment. With GPTs, anyone with $20/month can access the most advanced AI on earth, tweak it for their exact needs, and leverage the latest in LLMs, computer vision, image generation, and more to give them “AI superpowers”. Overnight, ChatGPT has gone from party trick to indispensable tool, no matter what you do.

• Build your personal customized ChatGPT by answering a series of simple questions and uploading documents

• Knowledge cut-off updated to 2023 plus real-time internet

• Multimodal: “sees” and “reads” uploaded files including PDF

• Create files, write code, reference knowledgebase of documents

• Interact with the world outside of the chat window using “actions”, turning it from a chatbot into an actual AI agent
  
  GPTs allows users to create true AI assistants for a variety of tasks. Whether it’s automatically drafting email replies, scheduling meetings, home and business process automation, providing financial analysis and more, users can configure sophisticated applications of these configurations, they are bounded only by their own creativity.

AI Unleashed: Unpacking Amii's Impact Report 2023
A big year lays the foundation for what’s ahead

What it is: The Alberta Machine Intelligence Institute (Amii), a globally-recognized leader in AI, has released its 2023 Impact Report. The report highlights key themes in the space in a year of extraordinary progress punctuated by groundbreaking applications like ChatGPT. AI Ethics, AI literacy, and the surge in AI-enabled startups have defined a year of innovation and rapid development.

What it means: Unsurprisingly, along with the surge of interest in AI research, there has been increasing adoption of AI tools by select industries. From altering business landscapes to influencing pop culture, AI's integration into daily life underscores the growing necessity for AI literacy, an area where Amii dedicates considerable effort building-out the province’s AI talent pipeline.

Why it matters: This year’s advances highlight a critical juncture in our society, with profound impacts on economy, culture, and ethics. This era of AI is redefining the boundaries of technology's role in addressing global challenges, necessitating a nuanced understanding of its ethical implications and its potential to democratize innovation across various sectors. Readers are invited to download the full report to explore these insights and understand the broader impact of AI as envisioned by Amii.

Leading Packs, Walling Gardens, Digging Moats, and Building Ecosystems
Plenty to unpack from OpenAI’s first Developer Day

What it is: OpenAI announced major enhancements to its services. The positives were somewhat tampered by a series of DDoS-driven outages later in the week.

• GPTs: Customizable “flavors” of ChatGPT designed for ease of use, allowing non-technical users to create personalized ChatGPTs. Critically, they can be integrated with external databases or knowledge bases, enhancing utility, scope, and accuracy.

• GPT-4 Turbo: A faster, cheaper version of GPT-4 with a quadrupled context window of 128k tokens, updated April ‘23 knowledge cut-off and new pricing structures​​.

• Assistants API: Enabling creation of "agent-like" experiences, for developers to build AI agents that can retrieve external knowledge or execute specific programming actions for diverse use cases.

• DALL-E 3 API: Now accessible via API with built-in moderation tools, extending creative possibilities in image generation with varied resolutions and formats

• Text-to-Speech API: Featuring six preset voices, enabling diverse vocal applications in AI-generated audio content.

• Copyright Shield: Protects businesses using OpenAI’s tools from IP claims, offering legal support (and fee payment) against lawsuits.

• GPT Store: The upcoming app store of GPTs to allow users to publish their custom GPTs in a dedicated store.

What it means: The breadth of announcements from OpenAI's Developer Day showcases the company's vision for the future of the space, one both wide and deep, but where it remains at the center. GPT-4 Turbo, with its significantly expanded context window is a critical enhancement, allowing for better in-session customization. The variety of APIs enable developers with critical tools to integrate AI into diverse applications that cross visual and audio modalities. Copyright Shield is significant in addressing the legal complexities and concerns businesses may have when integrating AI into their operations​, however we continue to believe that the better path is to train on data for which you actually have IP rights/permission.

Why it matters: Despite the “GPT Killers”, OpenAI remains in the top position. We see new models released each day (shout-out to Kai-Fu Lee’s Open Source Yi-34B model announced this week), and while they approach GPT-4 in some ways, none have the broad performance. No doubt, they will eventually catch-up to GPT-4, but OpenAI hasn’t been sitting still, it continues to train subsequent models and seems it will remain in the lead for the foreseeable future.

One more thing: As AI cements its role in our daily lives, its accessibility and reliability are paramount. OpenA’s recent outages underscore the importance of redundancy and backups in AI systems. Over-reliance on a single AI entity (especially a closed-source one) can be risky, emphasizing the need for a diverse, multi-faceted AI ecosystem to ensure continuous, reliable, and performant AI-enabled operations. Local, Open Source LLMs were a savior last week for those able to quickly switch during OpenAI’s turbulence.

Open Cerberus: Meta, Hugging Face, and Scaleway Forge Open Source Alliance
Fostering Innovation in France's AI Landscape

What it is: In a landmark collaboration, Meta (formerly Facebook), Hugging Face, and Scaleway have united to launch the "AI Startup Program" at Station F in Paris. This initiative, targeting the French entrepreneurial ecosystem, is a concerted effort to boost the adoption of open-source AI technologies​​​​. The program not only provides startups with mentorship and resources but also marks a strategic move by Meta to cultivate an open and collaborative approach to AI development in France​​. Notably, this partnership reflects a significant stance in the broader discourse on open versus closed AI development, with a focus on democratizing ethical AI​​.

What it means: The collaboration signifies a pivotal moment in AI development, especially within the open-source domain. Meta's alliance with Hugging Face, a champion of open-source AI, and Scaleway, a leader in European cloud infrastructure, exemplifies a strategic diversification away from the traditional tech giants like Microsoft and Google, often seen as proponents of a more closed AI approach​​. This initiative also highlights the growing importance of open-source AI as a means to ensure more transparent, accountable, and ethically guided AI development​​.

Why it matters: By fostering open-source AI development, this collaboration aims to drive innovation and ethical AI practices. The AI Startup Program at Station F stands as a testament to the potential of collaborative efforts in advancing AI technology, especially in terms of economic and technological benefits for the local ecosystem. Moreover, by supporting startups with resources and expertise, this initiative could catalyze a new wave of AI innovation in France, setting a precedent for similar programs globally​​​​. The focus on an open-source approach may also influence the broader AI development landscape, encouraging a shift towards greater transparency and ethical practices in AI.

Call John Ritter… Three’s Company
Duke joins Neuralink and Meta’s brain implant party

What it is: Duke University's latest brain implant joins the ranks of pioneering neurotech ventures like Elon Musk's Neuralink and Meta's mind-reading project (both previously covered by the AI Geekly). While each technology focuses on different brain functions—Neuralink aids muscle control, Meta decodes brain images, and Duke's device translates thoughts into words—they collectively push the boundaries of neurocommunication and AI integration​​​​.

What it means: This addition to the neurotech landscape, with Duke's device complementing the muscle control of Neuralink and the image decoding of Meta, signifies a quantum leap in harnessing brainpower for groundbreaking applications.

Why it matters: The convergence of these distinct technologies marks a new era in neuroscience and AI. Together, they promise a future where brain-computer interfaces address a wide array of human neurological functions and challenges.

Phoenix Ryzen: AMD's New APU Soars to New Heights
Ryzen 8000G APU Targets Entry-Level with AI Capabilities

What it is: AMD's upcoming Ryzen 8000G APUs, as hinted in the latest leaks, marks a notable update in its entry-level segment. This new series, integrating AMD’s Phoenix silicon and Ryzen AI accelerator, combines Zen 4, Zen 4c, and RDNA 3 architectures, bringing a combination of efficiency, performance, and AI capabilities to more budget-friendly hardware.

What it means: AMD's inclusion of AI accelerators in its entry-level APUs reflects the company's strategy to enhance the functionality of its more affordable product range, an area where it excels relative to Intel. This development suggests an emphasis on making advanced computing features, particularly AI, accessible within its broader hardware offerings.

Why it matters: By equipping its entry-level Ryzen 8000G APUs with AI accelerators, AMD is addressing a growing demand for AI capabilities in everyday computing. This update could attract a diverse group of users and developers, potentially leading to new applications and uses of AI in various settings (i.e. democratization). It's a noteworthy step for AMD, indicating their commitment to expanding the reach of AI technology within their product portfolio.

Everyone Has Their Sorrows
Nvidia can’t catch a break with China sanctions

What it is: Nvidia is adjusting its GPU strategy in response to US sanctions and market demands. They are expanding the availability of "China-Only" A800 GPUs to western markets due to US sanctions (which now prohibit their sale in China), while also introducing high-performance GPUs, like the H20, L20, and L2, in China.

What it means: Nvidia's actions highlight the challenges of international trade regulations and the importance of GPU technology in the tech industry. They are adapting to comply with sanctions, expanding their product range, and addressing the demand for GPU processing power in AI and other fields.

Why it matters: These developments impact the broader tech industry and AI specifically. With demand for AI GPUs at an all-time high, the eyes of the world are on every move of Nvidia silicon on the global scale. The introduction of competitive GPUs like the Huawei Ascend 910B in China adds to the AI hardware competition. This combined with the release of Kai-Fu Lee’s Open Source Yi-34B model this week should serve as a reminder to the US that China and other nations will not wait while the US regulates —the race has begun.

About the Author: Brodie Woods

With over 18 years of capital markets experience as a publishing equities analyst, an investment banker, a CTO, and an AI Strategist leading North American banks and boutiques, I bring a unique perspective to the AI Geekly. This viewpoint is informed by participation in two decades of capital market cycles from the front lines; publication of in-depth research for institutional audiences based on proprietary financial models; execution of hundreds of M&A and financing transactions; leadership roles in planning, implementing, and maintaining of the tech stack for a broker dealer; and, most recently, heading the AI strategy for the Capital Markets division of the eighth-largest commercial bank in North America.

Glossary

• GPTs: User-customizable flavors of OpenAI's ChatGPT platform that democratize AI capabilities, allowing non-technical users to create personalized chatbots with advanced features.

• LLMs: Language Models, such as ChatGPT, that learn to predict the next word in a sequence given the preceding words, fundamental in natural language processing tasks.

• Multimodal: AI capability to process and understand both text and images, including PDF documents.

• APU (Accelerated Processing Unit): A type of computer processor that combines CPU (Central Processing Unit) and GPU (Graphics Processing Unit) functionalities.

• Zen 4, Zen 4c, and RDNA 3 architectures: CPU (Zen) and GPU (RDNA) architectural designs developed by AMD for its processors and APUs.

• GPU: Graphics Processing Unit, a specialized electronic circuit used for rendering images and performing parallel processing tasks, crucial in AI for data processing.

• China-Only A800 GPUs: Graphics processing units designed specifically for the Chinese market due to trade sanctions.

• Huawei Ascend 910B: A competitive AI GPU developed by Huawei, contributing to the global AI hardware competition.

• Yi-34B model: An open-source AI model introduced by Kai-Fu Lee, offering alternative options in the AI landscape.

• Station F: a business incubator for startups, located in 13th arrondissement of Paris. It is noted as the world's largest startup facility.