The Sense of Business Leaders, as Seen in Fast Retailing Chairman Yanai | By Ken Kusunoki | Ten Minutes TV

The Sense of Business Leaders, as Seen in Fast Retailing Chairman Yanai | By Ken Kusunoki | Ten Minutes TV

When it comes to strategy, it's not just about combining different elements, but also considering the element of time. In fact, the depth and breadth of time play a crucial role in 80% of strategic planning. Those who can always say, "So what?" and create a narrative have a sense of business acumen. People with a sense of business can take any specific situation and abstract it into a logical framework by saying, "In essence, it means this." They keep that logic in mind and can apply it in any situation, which is a true sense of business acumen. This sense refers to the amplitude of the oscillation between concrete and abstract, the frequency of occurrence, and the speed of processing.

Google has set a new benchmark for AI language models (LLMs) with its latest model called PaLM. While the number of parameters is an important factor in LLMs, it doesn't always guarantee a better-performing model. PaLM 540B, with its 540 billion parameters, stands among some of the largest LLMs available, such as OpenAI's GPT-3 with 175 billion parameters, DeepMind's Gopher and Chinchilla with 280 billion and 70 billion parameters respectively, and Google's GLaM and LaMDA with 1.2 trillion and 137 billion parameters respectively. Microsoft and Nvidia's Megatron-Turing NLG also joins the league with 530 billion parameters.

Efficiency in the training process is a crucial aspect to consider when discussing LLMs or any other AI model. PaLM adopts a standard Transformer model architecture with some customizations. The Transformer architecture is commonly used in all LLMs, and while PaLM deviates from it in certain ways, the focus of the training dataset used is even more significant. PaLM's training dataset consists of a mixture of filtered multilingual web pages (27%), English books (13%), multilingual Wikipedia articles (4%), English news articles (1%), GitHub source code (5%), and multilingual social media conversations (50%). This dataset is based on the ones used to train LaMDA and GLaM. Nearly 78% of the sources are in English, with German and French sources accounting for 3.5% and 3.2% respectively, while all other sources have much smaller contributions.

In terms of performance, PaLM 540B has surpassed the few-shot performance of previous LLMs in 28 out of 29 tasks. It outperformed the prior top score achieved by fine-tuning GPT-3 with a training set of 7,500 problems and combining it with an external calculator and verifier, which reached 55%. PaLM's new score is approaching the average of 60% of problems solved by 9- to 12-year-olds, who are the target audience for the question set.

Combining the insights from both articles, we can draw a parallel between the sense of business leaders and the development of AI language models. Both require a deep understanding of the underlying principles and the ability to abstract complex situations into a logical framework. The sense of business acumen allows leaders to navigate through challenges and make strategic decisions based on a comprehensive understanding of various factors. Similarly, the development of AI language models involves training the model on diverse datasets and customizing the architecture to optimize performance.

Incorporating unique insights, we can see that the sense of business acumen and the development of AI language models share a common characteristic of balancing between concrete and abstract thinking. Successful business leaders possess the ability to abstract complex situations into simple and logical frameworks, allowing them to make informed decisions. Similarly, AI language models like PaLM are designed to understand and generate human-like language by abstracting patterns and structures from vast amounts of training data.

With these commonalities in mind, here are three actionable pieces of advice for business leaders and AI researchers alike:

1. Cultivate a sense of abstraction: Develop the ability to extract the core essence of complex situations and think in abstract terms. This allows for better decision-making and problem-solving.

2. Embrace diversity in training data: Just as diversity in business perspectives leads to comprehensive decision-making, training AI language models on diverse datasets can enhance their understanding and performance.

3. Continuously adapt and customize: Business leaders should adapt their strategies to the ever-changing market dynamics, while AI researchers should customize their models and architectures to optimize performance.

In conclusion, the sense of business acumen demonstrated by leaders like Fast Retailing Chairman Yanai and the advancements in AI language models like Google's PaLM share common traits of abstraction and customization. Both require a deep understanding of underlying principles, the ability to process complex information, and the agility to adapt to changing circumstances. By incorporating these traits and actionable advice, business leaders and AI researchers can continue to drive innovation and make informed decisions in their respective fields.