Topics In Demand
Notification
New

No notification found.

Unlocking AI's Power : Enterprises Must Crack the Data Code
Unlocking AI's Power : Enterprises Must Crack the Data Code

February 20, 2024

120

0

 

A transformative force

Generative AI (Gen AI) is causing quite a stir—across industries and individuals alike. Estimates suggest that it could add up to $2.6 Tn to 4.4 Tn annually to global economy[1], comparable to one of the top 5 largest world economies.

Of the three distinct components of Gen AI – data, algorithms and computing, the latter two have significantly been taken care of, however, data remains a complex challenge to deal with.

This is due to another mega trend– Data explosion! Data has been growing at an unfathomable trajectory, expanding 100,000 to 150,000 times, over the last 25 years, incomparable to any development in human history.

However, as organizations attempt to get their hands around all this ever increasing, dynamic data, it becomes an overwhelming, mammoth task, and at the same time they are ineffective with the data they already have. This is called the Data Paradox – where amidst the deluge of data, actionable insights remain scarce for organizations, similar to the dilemma of “water, water everywhere, nor any drop to drink”.

Transformative value generation with Gen AI

Despite the huge potential of Gen AI, there is a significant gap between the technological advancement and the business reality. Bridging this gap primarily hinges on refining and effectively leveraging data. The foundational models of Gen AI that harnesses the macro data, the vast internet data, are enabling us to get to 70-80% of the answers we are typically looking for. This is undoubtedly a great starting point but provides only 20% value. To realize transformative value, any model must solve specific business problems. This requires the foundational models to be meshed with proprietary enterprise and individual data for effective contextualization. Connecting the three layers is the key to realize 80% value.

With Gen AI, the data paradox becomes even more stark as the magnitude of data becomes too big. Specifically, there are 4 data challenges to be solved to unlock the value of Gen AI::

1. Data contextualization challenge

For the effectiveness of Gen AI, organizations must identify and combine the right data. Mapping data across the three layers – the macro, the enterprise and the individual, is a complex task, especially due to the dynamic nature of data, scattered across different systems.

 

2. Data integration becoming even more complex
Organizations must integrate the externally available large language models data with internal proprietary data. Organizations are already dealing with highly diverse and dynamic datasets that pose integration challenges. Integrating these effectively with the LLM models that leverages the data of the world, adds greater complexity to the whole exercise.

3. Newer challenges to Data Quality

Maintaining data quality has been increasingly difficult in the big data world owing to the increased heterogeneity of data. Gen AI models add to the challenge as skewed input data may bias the output. Gaps in data, less or inaccurate data may lead to “hallucinations” – producing inaccurate information as if correct. Additionally, capturing the context for highly subjective unstructured data is a challenge.

 

4. Data Security, a grave issue

Foundational models of Generative AI typically operate on an open-loop architecture, residing in third-party environments. This exposes the organization’s proprietary data to bigger risks like confidentiality breaches, compromising proprietary information, non-compliance with regulations, and others. In my experience, this is the biggest issue that is limiting enterprises in moving Gen AI use cases from labs to production.

 

In conclusion, the promise of Gen AI remains immense, however, unleashing its full potential hinges on overcoming key challenges, particularly solving the data paradox. 

 

Author: Nitin Seth, Co-Founder & CEO, Incedo Inc. 
 

 


That the contents of third-party articles/blogs published here on the website, and the interpretation of all information in the article/blogs such as data, maps, numbers, opinions etc. displayed in the article/blogs and views or the opinions expressed within the content are solely of the author's; and do not reflect the opinions and beliefs of NASSCOM or its affiliates in any manner. NASSCOM does not take any liability w.r.t. content in any manner and will not be liable in any manner whatsoever for any kind of liability arising out of any act, error or omission. The contents of third-party article/blogs published, are provided solely as convenience; and the presence of these articles/blogs should not, under any circumstances, be considered as an endorsement of the contents by NASSCOM in any manner; and if you chose to access these articles/blogs , you do so at your own risk.


© Copyright nasscom. All Rights Reserved.