Friday is back and so is another edition of “Explore the AI Landscape”!

Before diving in, we want to take a minute to thank all of of our subscribers for your support. Without you, this Newsletter would be pointless. With that in mind, we are hoping some of you could help us out with giving some feedback, something we here at TDAI thrive on. If you are inclined and have 10-60 seconds, it would be awesome if you could let us know if the content here is impactful and what else we can add that may add to your delight. Taking the poll at the bottom or sending a message would mean the world. Whether it’s telling us how we can improve, new content ideas or anything else you want to share, we would love to hear from you!

Let’s get into it:

Last week, ChatGPT and GPT-4 were put to the test, quite literally, and administered the CFA Level I and Level II exams…it didn’t go well. In other news, the U.S. Tightened China’s Access to Advanced Chips for Artificial Intelligence, Microsoft started a bug bounty program, and the “NO FAKES Act of 2023” was released. Quite an interesting week.

In this edition:

The rise of large language models (LLMs) such as ChatGPT and GPT-4 has been nothing short of revolutionary. Their prowess in understanding and generating human-like text has led to speculation about their potential in more specialized domains. One question that has been on the minds of many is whether these models could tackle and pass the notoriously challenging Chartered Financial Analyst (CFA) exams. (If you want to see the actual data sheet and results, they can be found here)

The CFA Exam: A Brief Overview

The CFA Program is a globally recognized certification for finance professionals. It comprises three levels of exams that delve into topics like financial analysis, portfolio management, accounting, and economics. The exams are renowned for their rigor, testing not just knowledge but also practical financial reasoning. With pass rates hovering around 40-50%, the CFA is not for the faint-hearted.

The Experiment

Researchers from Queen's University, Virginia Tech, and J.P. Morgan decided to put ChatGPT and GPT-4 to the test. They administered mock CFA Level I and Level II exams to these models. The study's methodology was comprehensive, evaluating the models in zero-shot, few-shot, and chain-of-thought prompting scenarios.

Key Findings

Based on the study's criteria, only GPT-4, with the aid of few-shot prompting, could potentially pass the exams. However, it's a close call, especially for Level II.

Reflections

The results of this study are both promising and revealing. They underscore the incredible advancements in AI, with models like GPT-4 coming close to passing professional exams. However, they also highlight the nuances and depth of human expertise. The CFA exams test not just knowledge but the application of that knowledge in complex scenarios. While LLMs like GPT-4 show promise, they still have some way to go.

Moreover, the study's findings on few-shot prompting and chain-of-thought reasoning offer valuable insights. They suggest that while these models have vast inherent knowledge, they can benefit significantly from context and guidance.

Looking Ahead

The journey of AI in specialized domains like finance is just beginning. With targeted training and more advanced methodologies, it's conceivable that future iterations of these models could perform even better. However, for now, finance professionals can take solace in the fact that their expertise remains unparalleled.

In the broader context, this study is a testament to the potential of AI. It challenges us to think about the future roles of humans and machines in professional domains. While AI might come close to human performance in certain areas, the synergy of human-AI collaboration might be the real game-changer.

The global tech landscape is witnessing a significant shift as geopolitical tensions manifest in the realm of technology. A prime example is the ongoing tussle between the U.S. and China over advanced technologies, particularly in the domain of artificial intelligence (AI).

The Nvidia Ban: A Recap

In August of the last year, the U.S. imposed a ban on Nvidia's chip exports to China. This move jolted China's emerging AI startups, pushing them to seek alternative solutions. Nvidia, in an attempt to navigate around these restrictions, introduced chips with toned-down performance. However, this workaround was short-lived.

This week, the Biden administration unveiled a series of measures aimed at curbing Beijing's military aspirations. Central to these measures is a further clampdown on Nvidia's AI chip exports to China. Specifically, the A800 and H800 chips, which Nvidia had specially designed to maintain shipments to China, are now under the scanner.

While the primary intent behind these bans is to limit China's military capabilities, the ripple effects are felt most acutely by China's burgeoning AI startups. Major tech giants like Alibaba, Baidu, ByteDance, and Tencent had collectively placed orders worth $5 billion for the A800 chips for the current and upcoming year.

The Broader Implications

Despite these setbacks, Chinese tech behemoths remain undeterred in their AI pursuits. Baidu, for instance, recently launched its flagship foundation model, Ernie 4.0, boasting capabilities on par with GPT4.

However, the challenges for AI companies in China are multifaceted. Apart from grappling with chip shortages, they also have to navigate stringent regulatory requirements. Beijing mandates that LLM-based services secure a license before catering to the public. This necessitates strong government relations and adeptness in maneuvering bureaucratic hurdles.

Interestingly, these challenges have inadvertently paved the way for generative business intelligence services to thrive. These services, which demand lesser computational power and are easier to regulate, are becoming increasingly popular.

Final Thoughts

The U.S.-China tech standoff underscores the intricate interplay between geopolitics and technology. While bans and restrictions might seem like immediate setbacks, they often lead to innovation and alternative solutions. However, as the tech world becomes more polarized, collaboration and shared progress might become casualties.

E-commerce has transformed the way we shop, and with the integration of Artificial Intelligence (AI), the online shopping experience is set to become even more personalized, efficient, and user-friendly. As the demand for online shopping continues to surge, AI tools are stepping up to offer solutions that enhance product recommendations, streamline customer service through chatbots, and optimize the supply chain. In this collection, we delve into some of the top AI tools that are making waves in the e-commerce sector.

Explore the Collection

Own a tool that you would like to see on our site or newsletter please Submit a tool or reach us by replying to this email or at [email protected]

Prompt: Deity woven of sunlight and rain. Gold and blue. Masterpiece.

Prompt: Cosmic God looking over the mountains fantasy art