Your AI & ML Initiatives will FAIL unless…

AI & Machine Learning Initiatives are not delivering results for many enterprises:

AI and Machine Learning are accelerating success of Facebook, Amazon, Apple. And because of their success, many enterprises started investing in transformative initiatives using AI & Machine Learning technologies. But their investment in AI & ML are not yielding desirable benefits! Why? The quality of insights that AI & ML algorithm churn out depends on the completeness of the input data. AI & ML initiatives of Facebook, Amazon, Apple, Google are successful because they collect huge amount of data about their customers. Their data completeness factor is quite good; hence their AI & ML churn out differentiating insights.

But most of the enterprises collect only small amount of data about their customers, their preferences, life-style, product usage & experience. And in absence of rich data, best of AI & ML Algorithms fail to churn out disruptive insights. To product disruptive insights, AI & ML would need data beyond the enterprise. The Enterprise Data needs to be enriched with the data available with the partners. Enterprise eco-system data need to be collected before any AI & ML initiative could detect hidden patterns or insight provoking correlations.

What need to be done to improve Data Completeness

Enterprise need to acknowledge that their data is siloed. Unless it is enriched with the data of their partners in the eco-system, the data would not be of use to generate disruptive insights. Enterprise need to de-centralize their control of their data. And once they do it, their partners will be able to share their data with the enterprise — — provided trust is ensured using digital mechanisms and technologies.

How the TO-BE Architecture Looks like:

In the TO-BE Architecture, the Data is made comprehensive by enriching the enterprise data with the partners data on which the AI & Machine Learning algorithms are applied. Once data is complete and rich, you start getting disruptive insights from the same AI & Machine Learning algorithms. The TO-BE architecture looks like the following:

How to Transition from “AS-IS” to “TO-BE” Architecture:

The transitioning from ‘AS-IS’ to ‘TO-BE’ requires Enterprise and its Partners to make the data and algorithms portable.

In the TO-BE Architecture, the Data and Processing Algorithm is decoupled and both are portable so that they can be shared.

Architecture and Technology Options for the “TO-BE” Architecture:

You need to choose Decentralized Architectural Concepts which can be summed up in 4 bullet points:

1. DECENTRALIZED PROCESSING: It means no single entity controls the processing application. Algorithms are portable and can be executed securely anywhere.

Blockchain platforms should be chosen to realize above capability. Ethereum is one of the most popular choice in open source. IBM Blockchain based on Hyperledger is another candidate choice.

2. DECENTRALIZED FILE PERSISTENCE: It means no single entity controls the persistence of large data files.

Decentralized File Management platforms like IPFS should be chosen for this.

3. DECENTRALIZED DATABASE: It means no single entity controls the persistence of data in database.

Blockchain Database platform like ‘BigchainDB’ should be chosen for this. Almost all of the Blockchain platforms provide database. But they are struggling to provide scalability. So, I suggest you to evaluate ‘BigchainDB’for huge scalability needs.

4. IMMUTABLE: It means no one can change the data and the algorithm. It’s temper proof.

It is because of above capabilities, enterprise and their partners get the trust for sharing their data.

Conclusion:

Data is the new oil. AI and Machine Learning can process it and churn disruptive insights for your enterprise if the underlying data is rich and complete. It can be done if partners share their data with the enterprise. It will happen only if partner get trust that their data will not get misused. They will remain owner of their piece of data and it will be used with their concurrence only. But trust will not be based in personal relationships. It will be based on modern digital technologies leveraging decentralized architecture & associated technologies.

*** End of Document ***