AI and the Changing Role of Data

The direction of AI development has evolved from merely executing simple rules to handling increasingly complex tasks. As inputs became more complex (and more unstructured), more sophisticated rules were required to understand and process them. This led to the adoption of machine learning, but it also introduced new bottlenecks. Machine learning, which required massive infrastructure and highly skilled personnel, ultimately restricted advanced AI to specific industries or large corporations.

Is Data No Longer Important?

With the advent of AI democratization through LLMs/Cloud AI, there has been a significant shift in how companies utilize data. In the past, securing data and training AI models on it was critical, but now we are entering a new era where AI's inference capabilities can be leveraged. Previously, it was common for each company to develop its own AI, but now it seems that centralized, large-scale development and operation of AI models will become the norm. Just as electricity is generated in power plants and used by households, and just as companies once ran their own data centers but have now shifted to on-demand Cloud computing, companies can now utilize pre-trained AI inference capabilities as needed. As a result, the importance of 'training' AI is expected to diminish.

The Importance of 'Context/Sensing' Data and LLM Context Window

Of course, data itself remains important. However, rather than using data to 'train' AI models, the focus is shifting to implementing AI functionalities that were previously impossible due to the complexity of development and operations. Now, 'sensing' data, which provides the current context to AI—especially in customer-facing services and products—will become increasingly important. This data may not necessarily need to be collected and stored by the company itself.

Attention should be paid to the growing LLM Context Window. LLMs can process increasingly large amounts of input data, which now includes not just text, but also images, videos, and other forms of data. This will significantly enhance AI's ability to rapidly acquire and process information.

What Companies Should Focus on Moving Forward

Companies should now refocus on implementing AI functionalities that they were previously unable to pursue due to cost and complexity. Providing AI with context data and sensing data is becoming increasingly important, opening up new opportunities to deliver innovative customer experiences.

In the end, while data remains important in the era of AI advancement, its purpose and role are changing from the past. Companies should now focus less on data collection and processing, and more on providing the appropriate inputs that enable AI to produce optimal results. By leveraging this contextual data, they can create new experiences in customer-facing services and products.