Topics In Demand
Notification
New

No notification found.

Generative AI Running into Breaking Points While Breaking Through
Generative AI Running into Breaking Points While Breaking Through

November 19, 2024

AI

134

1

 

Author: Ankit Rana, CTO, Polestar Solutions 

 

There must have been rarely a conversation without the mention of AI or Generative AI in the past two years – especially within the leadership & tech community. It has been about where we can get Generative AI into operations or what’s the best way to get started. But now that the initial hype has toned down – let’s see what’s working and what’s not!  

Generative AI has added a lot of value, there’s no doubt about it. There are wins including – automating tasks, personalizing interactions, easier data communication, and quicker resolutions. But there are also issues like bias, hallucinations, handling the logical capabilities of Gen AI, the choices between multiple LLMs, etc. So, the question comes back to how truly progressive are we in Gen AI adoption? 

The truth behind the adoption 

The reality is more nuanced than what the headlines suggest. While the Gen AI adoption in companies ranges from 75% to 85% (based on various reports), our experience shows that most of them are stuck in the pilot or PoC stage or have a limited scope of implementation. On the contrary, it has shown an increase in personal productivity especially when augmented with human intelligence.  

FOMO was never a great way to get started with things – because it points to the things that are lacking inherently: 

  1. Infrastructure readiness: To truly leverage the Generative AI capabilities, you need to move out of legacy systems, data silos, and inconsistent data quality.  

  1. Skills deficit: There's a significant shortage of professionals who understand both the technical aspects of Gen AI and its business applications. You’ll need more folks who understand the nuances of working with AI.  

  1. Cost reality: When calling for APIs for any tool be it OpenAI or GPT or AWS – it is usually measured in tokens, or the maintenance and model training is quite underestimated.  

In fact, 63% of customer experience (CX) leaders report that the overall investment required to implement AI has been higher than expected, highlighting the gap between expectations and reality​. Plus, to support the level of capabilities companies want building the infrastructure including data centres, utilities, and applications would need roughly a $1Tn in the next several years according to Goldman Sachs. 

Not everything is bleak: Where is it taking a turn 

Though some might disagree, we’ll still believe that AI is the next point of transformation. The whole discussion should be around the why and the how. One thing Generative AI has done is to steer the conversation towards AI at an exceptionally rapid pace. (Everyone just wanted to get started with it).  

So, this led to an influx of use cases being realized - majorly for the productivity improvement. One great example of this is how we integrated code generation capabilities with our MDM solution – which gives the appropriate code based on the parameters and the data pipeline to be created. We estimate that tasks that previously required hours can now be completed in just 5–10 minutes – saving hours of redundant time. 

Similarly, some of the key use cases we’ve seen that are working well are: 

  • Code Development: AI has been successful in automating low-level code writing, which allows developers to focus on more complex and productive tasks. 

  • Creative Design: Generative AI can generate design ideas in minutes that would have previously taken hours, speeding up the process of bringing new ideas to market (not to mention creating hypothetical images based on text) 

  • Customer Support: ServiceNow has reported an 80% decrease in the average time it takes to resolve customer service issues using AI  

  • By using LLMs for test generation, engineers can create more tests in less time, leading to improved code quality and fewer bugs. 

  • Data retrieval and querying applications like P. AI which can integrate into existing workflows like Teams to ask questions about data for easier analysis (without being a technical expert)  

For a few this might be the low hanging fruits – for others it might be value adding. You might argue that there are other tools like NLP, etc. which could have done a few of these but the time taken to do it would have been considerably longer. 

Need for an inclusive approach: Build for the future not due to FOMO 

This is just the start. In the process of building value adding and ‘thick-wrapper’ Gen AI solutions – there must be change in the whole approach and the process. And this path needs to acknowledge both the potential and limitations of Gen AI. Success will come not from rushing to adopt every new AI capability, but from thoughtfully integrating these technologies in ways that create sustainable value for all stakeholders. 

The facets are multi-fold. From identifying the clarity behind multiple models Eg. transformer models like Claude, GPT4 or diffusion models like Midjourney and DALL-E etc. and their licensing rules. To the way you query, prompt engineer or tips are to make things easier and feasible – given that logical reasoning is still not one of the greatest qualities of Gen AI.  

And then the question of where to get started. We propose having a right value vs performance feasibility matrix before getting started – to identify the low hanging fruits which provide best value. Also keep in mind that Generative AI is still AI – the nuances and the background you lay for it should still be solid enough. Therefore, ensure that you have the right data engineering practices like data lakehouse or right data ecosystem like data fabric to support your journey.  

Also, don’t be afraid to experiment. Just know when to stop! 

### 

 

 


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.


About Polestar Insights Inc. Founded in 2012 by Chetan Alsisaria (CEO & Co-Founder), Amit Alsisaria (COO & Co-Founder) and Ajay Goenka (CFO & Co-Founder), Polestar, is a leading AI & analytics solutions company that serves Fortune 1000 companies, startups and the government across various industries, including CPG & retail, manufacturing and pharmaceuticals, among others. Headquartered in Dallas, Texas, the company enables businesses across North Americas, Asia Pacific, ANZ, and the UK with analytics foundation, data science and AI initiatives, offering a comprehensive range of services to help succeed with their data. For more information visit: https://www.polestarllp.com/

© Copyright nasscom. All Rights Reserved.