From Model-Centric to Data-Centric – ML’s Evolution

By now, you might be thinking: if data-centricity is essential to the further evolution of AI and ML, how come model-centricity is the dominant approach?

This is a very relevant question to ask, and one we will answer in this chapter. To understand what it takes to shift to a data-centric approach, we must understand the forces that have led to model-centricity being the predominant approach, and how to overcome them.

We will start this chapter by exploring why the evolution of AI and ML has predominately followed a model-centric approach, before diving into the huge opportunity that can be unlocked through data-centricity.

Throughout this chapter, we will challenge the notion that ML requires big datasets and that more data is always better. There is a long tail of small data ML use cases that open up when we shift our mindset from bigger data to better data.

By the end of this chapter, you will have a clear...

Exploring why ML development ended up being mostly model-centric

A short history lesson is in order to truly appreciate why a data-centric approach is the key to unlocking the full potential of ML.

The fields of data science and ML have achieved significant advancements since the earliest attempts to make electronic computers act intelligently. The intelligent tasks performed by most smartphones today were nearly unimaginable at the turn of the 21st century. Moreover, we are producing more data every single day than was created from the beginning of human civilization to the 21st century – and we’re doing so at an estimated growth rate of 23% per annum1.

Despite these incredible developments in technology and data volumes, some elements of data science are very old. Statistics and data analysis have been in use for centuries and the mathematical components of today’s ML models were mostly developed long before the advent of digital computers.

For our purposes...

Unlocking the opportunity for small data ML

The group of tech companies famously labeled The Big Nine by author Amy Webb18 are examples of consumer internet companies that have leveraged big data and AI to build world dominance. Amazon, Apple, Alibaba, Baidu, Meta, Google, IBM, Microsoft, and Tencent dominate in the digital era because they utilize enormous amounts of user data to power their AI systems.

As network-based AI-first businesses, they have amassed customers on an unprecedented scale because users are happy to co-create and share their data, so long as it is a net benefit to them. For the Big Nine, getting enough modeling data is rarely a problem, and investing in the most advanced ML capabilities is a virtuous circle that enables more market dominance.

For most other organizations – and ML use cases – this sort of scale is unachievable. As we explored in Chapter 1, Exploring Data-Centric Machine Learning the long tail of ML opportunities doesn’...

Why we need data-centric AI more than ever

The leading organizations in AI, such as the Big Nine, have achieved incredible results with ML since the turn of the century, but how is AI being used in the long tail?

A 2020 survey published by MIT Sloan Management Review and Boston Consulting Group concluded that most companies struggle to turn their vision for AI into reality. In a survey of over 3,000 business leaders from 29 industries in 112 countries, 70% of respondents understood how AI can generate business value and 57% had piloted or productionized AI solutions. However, only 1 in 10 had been able to generate significant financial benefits with AI.20

The survey authors found that companies that were realizing significant financial benefits with AI had built their success on two pillars:

• They had a solid foundation of the right data, technology, and talent.
• They had defined several effective ways for humans and AI to work and learn together. In other words, they...

Summary

In this chapter, we reviewed the history of ML to give us a clear understanding of why model-centric ML is the dominant approach today. We also learned how a model-centric approach limits us from unlocking the potential value tied up in the long tale of ML opportunities.

By now, you should have a strong appreciation for why data-centricity is needed for the discipline of ML to achieve its full potential but also recognize that it will require substantial effort to make the shift. To become an effective data-centric ML practitioner, old habits must be broken and new ones formed.

Now, it’s time to start exploring the tools and techniques to make that shift. In the next chapter, we will discuss the principles of data-centric ML and the techniques and approaches associated with each principle.

References

1. https://www.idc.com/getdoc.jsp?containerId=prUS47560321, viewed on 23 September 2022
2. Lewis, A. R., 2006, The American Culture of War, Routledge, New York, USA
3. https://www.nasa.gov/feature/when-the-computer-wore-a-skirt-langley-s-computers-1935-1970, viewed on 23 September 2022
4. https://www.historyofdatascience.com/k-nearest-neighbors-algorithm-classification-and-regression-star/, viewed on 23 September 2022
5. http://large.stanford.edu/courses/2012/ph250/lee1/docs/Excepts_A_Conversation_with_Gordon_Moore.pdf, viewed on 23 September 2022
6. https://www.businessinsider.com/ibm-1970-mainframe-specs-are-ridiculous-today-2014-5, viewed on 22 September 2022
7. https://www.ibm.com/ibm/history/exhibits/mainframe/mainframe_PP3145.html, viewed on 22 September 2022
8. https://www.apple.com/au/iphone-14/specs/, viewed on 23 September 2022
9. https://www.quickbase.com/articles/timeline-of-database-history, viewed on 24 September 2022
10. https://www.dataversity.net...