Activities
Divisions
Programs
Activities
Divisions
Programs
Special Feature: Amit Luthra, Managing Director (ISG), Lenovo, discusses the importance of Independent Data for Autonomous AI
Approximately 93% of Indian corporations plan to fund AI in 2024, not merely focusing on GenAI and LLMS, but also on AI inference models. In this context, creating Autonomous AI is vital, but it will be tremendously challenging without Independent Data.
While artificial intelligence has already shown remarkable progress, we've only scratched the surface of its potential. Much of our attention has been directed towards generative AI and LLMs, however, advancements in other areas are happening rapidly.
For instance, 93% of Indian corporations are intending to invest in Artificial Intelligence (AI) by 2024. They are not only interested in General AI and Learning Management Systems but also in AI inferencing models that assist them in examining and understanding scenarios, as well as predicting appropriate responses. Despite General AI creating a mixture of admiration and fear, Indian companies see the future in analytical AI, automation, edge-computing, and AI inference.
We had a discussion with Amit Luthra, the Infrastructure Solutions Group's Managing Director at Lenovo India, during the launch of their 'CIO Playbook 2024 – It's all about Smarter AI' report. We talked about the future of AI in India.
We cover everything, ranging from data autonomy and ethical principles regarding AI, to the misconception of AI "taking" our employment and how CIOs intend to utilize AI, not merely because it's a trending term, but due to its potential to make a significant difference. Below are the abridged segments from our discussion:
What changes have occurred in India's technological requirements over time, particularly in relation to AI? India's digital progression was built on four key areas – Social Media, Mobility, Analytics, and Cloud Computing. Analytics always incorporated an element of AI. Most business applications have attempted to integrate Artificial Intelligence. However, two significant developments completely altered the landscape.
Initially, as the Internet of Things (IoT) became more widespread, a large quantity of data started to be collected from various sensors. Businesses understood that they couldn't afford to wait to collect all the data and then analyze it afterwards to achieve their objectives. For instance, if a manufacturing plant wanted to boost the quality of their output, they needed to scrutinize all their data in real-time, especially if the data was time-critical. This drove up the demand for Edge Computing.
The next step involves the onset of GenAI and the construction of substantial LLMs. To create these extensive LLMs, you require advanced systems with significant computational capacity.
Nonetheless, I believe that presently in India, a significant portion of investments is concentrated on developing LLMs. However, moving forward, we will also need to emphasize on enhancing investments in AI inferencing. The growth of inferencing is expected to be much faster. This will require a combination of CPU-GPU workloads, rather than solely depending on GPU-based systems.
What are the main difficulties businesses, particularly new ones, face when implementing AI? AI, particularly generative AI, requires a well-organized structure. The primary difficulty lies in establishing robust data management and ensuring the accuracy of data. If you input low-quality data into your AI models, you'll get low-quality results. You might develop an impressive LLM that learns rapidly and has remarkable algorithms, but if the training data is incorrect, incomplete, out-of-date, or poor quality, it could lead to problems.
It's equally crucial to establish an analytical structure that aligns with specific objectives. Businesses must understand the purpose behind their need for an AI model and what they aspire to accomplish. They should question, "What is my end goal with the collected data?"
Many companies' AI initiatives fail due to a lack of clarity on the purpose of utilizing AI. Some firms might aim to enhance customer happiness, while others might aim to boost manufacturing.
Some might wish to establish cleanliness procedures to enhance efficiency, whereas others might just want to deter theft and scrutinize fraud. It's essential for companies to be certain of their ultimate goal and what they intend to accomplish. Once this is understood, the appropriate technology can be implemented seamlessly.
Next, we must consider both the safety and moral aspects of this matter, giving each equal weight. When handling data, it can either elevate or ruin you. Companies must consider how to tackle the moral dimension of this issue, and where boundaries should be set. Additionally, they must ensure that they manage confidential data properly, utilizing an appropriate level of care and a suitable security structure.
The CIO Playbook 2024 reveals that almost half of CIOs face difficulties when trying to fill AI-related positions. Despite India having the most substantial number of engineers and developers, it seems there might be a gap between the industry's expectations and our engineers' existing skills. Although AI has been a part of the engineering curriculum for over 25 years, the transition from academic learning to practical application in the modern corporate environment is not seamless. It is unrealistic to assume that a newly graduated student can immediately demonstrate proficiency in AI work.
The issue isn't a lack of resources, but rather creating a supportive structure for these resources.
This can be especially difficult as circumstances often evolve rapidly. What was deemed essential last year might be irrelevant this year. A business leader might currently be wondering how they can constantly mentor and educate their teams, or how they can steadily foster a structure for learning.
The problem of retention then comes into play. The act of poaching is a significant concern, particularly in the context of AI. Under these circumstances, companies must consistently pump funds into their resources, and they must also possess the determination to keep these resources.
Discussions about Sovereign AI and Data Sovereignty are rampant nowadays. Given the current development of LLMs, what potential obstacles do you foresee in establishing sovereignty? Data sovereignty involves understanding the exact location of your data and making sure it is stored within the geographical confines of your own nation. Generally, data is saved on the cloud, but the physical hosting of the cloud could be in a different country.
For instance, numerous multinational corporations previously operated centralized data centers situated anywhere globally for more efficient and streamlined processes. In such a scenario, you cannot achieve sovereignty as the data, potentially containing personal details of Indian citizens, is stored outside the nation.
In this context, the importance of data sovereignty legislation cannot be overstated. It stipulates that personal data of Indian citizens, for instance, must remain within the country's borders.
After confirming this, it's the responsibility of the security and analytics framework, as well as the government, to step in. These frameworks must guarantee that crucial information is not being exposed, and is managed, stored, and adhered to according to the specified rules.
So, how do Indian companies utilize this information and build an analytics structure around it? It's crucial to understand what you aim to accomplish and the goal you're working towards. Each sector and application is unique.
What types of LLMs or applications might we expect in this situation? Allow me to illustrate with an example. If I need to contact the customer support team of a service I'm using, I would rather call them directly, instead of interacting with a chatbot through an app, primarily due to often unsatisfactory experiences. However, we are now witnessing the emergence of AI-powered voice assistants serving as customer service representatives, with all interactions being voice-based.
A further instance could be in the retail sector. In the near future, we might see digital representations replacing traditional sales staff, designed to address your unique questions according to your specific needs. Imagine you walk into a dealership and request a sales clerk to break down the details of a new vehicle. A less seasoned salesperson might create uncertainty for the buyer. Additionally, there's a possibility they may not have the answers to your inquiries if you pose a particularly specific question, which although related to the product, is seldom asked.
Consider the same situation, but instead there's a virtual representation almost identical to a human and is highly knowledgeable about its assigned product. After only a single training session, it can address all your inquiries.
Furthermore, the company doesn't have to invest considerable time in retraining their employees for any new updates. During this procedure, the only task is to verify if the language models, regardless of their size, are acquiring knowledge in real-time and have the most recent data sets at their disposal. Think about how distinct the previous versions of ChatGPT are compared to the GPT4.5-driven ChatGPT.
After the construction of the LLM, it is crucial to ensure that it is transferable, current and, most importantly, simple to implement. This is where Lenovo steps in to assist businesses with a variety of solutions, extending from handheld devices to cloud-based systems. This is because AI is not merely confined to a single LLM, but rather, it spans the whole ecosystem.
India's government hasn't yet established rules pertaining to AI in the country. What kind of regulations would you prefer from a business standpoint? Additionally, how do you think the Digital Personal Data Protection Act of 2023 will affect the progress of AI in India? It's essential to have a robust framework for security and ethics. How can we protect people from the deepfakes that are already in circulation? The issue of security and ethics is going to be the main topic in this context. Once the situation is fully understood, companies developing LLMs will be able to set boundaries effectively.
Top-level executives, regardless of their level of expertise in AI or technology overall, are enthusiastic about incorporating AI into their companies. So, what's the best way for Chief Information Officers to engage them and guarantee that AI is deployed appropriately and in the right areas? As our research for the CIO Playbook 2024 has shown, the desires of the business and the demands of technology are not always aligned. Many high-ranking executives are eager to focus on generative AI, possibly because it's a popular trend.
In 2023, generative AI was not a focus, but now it takes precedence as it has turned into a trending topic. Last year, the primary discussion was around customer satisfaction and experience. However, currently, it has become their second most important concern as everyone understands that attracting a new customer is costly, while keeping an existing one is far more cost-effective.
Likewise, income and earnings expansion, which has always been the top concern for all companies, has now dropped to third place. Sustainability, which was previously not given much attention, is now the fifth most important priority due to ethical considerations.
Let's now examine what the tech department, particularly the Chief Information Officer (CIO), considers as their main focus. CIOs consistently view automation as their top concern. By giving precedence to automation, numerous issues can be addressed. Therefore, artificial intelligence, particularly generative AI, becomes a consideration only after establishing automation.
It's essential to establish a specially designed infrastructure for high-performance tasks related to AI. Prioritizing security and ethical considerations should be fundamental to your AI development. As I've mentioned before, all of these components serve as a structure. AI can't function autonomously, it needs to serve a specific purpose. The application of AI is akin to a strategic plan, with various key elements, all supported by a guiding framework. We've always been advocates of this approach, and it's gratifying to see it validated in this survey when we revisit and discuss with CIOs.
What sectors should be prioritized by the government when integrating AI into their administrative tasks? In what ways can AI improve government operations? The government has established a clear direction regarding AI. As citizens who have voted for this government, it's essential we align with their plans. When government officials and experts interact with us – the industry leaders, at various events, they collect diverse opinions and insights on different topics. They also encourage us to offer our viewpoints and suggestions for ideal practices.
In my opinion, when the government develops a clear understanding of AI and establishes a structure akin to data sovereignty, it will be mandatory for everyone to adhere to it. Furthermore, our Prime Minister, Shri Narendra Modi, has also endorsed the concept of 'AI for All'. We at Lenovo have been promoting 'AI for All' for a considerable amount of time. We are of the belief that if we can achieve AI for all, it will strengthen our vision immensely.
Ultimately, the burning question everyone is pondering is: How will AI transform the current landscape of work and jobs? AI is simply an additional technology or toolset that will assist people in improving their performance. Take for example the soldiers in World War I, World War II, and all other wars that have been fought. During World War I, soldiers were equipped with .303 rifles, whereas now they have LMGs and MMGs. Despite the drastic evolution of technology and change in tools, there is still a need for soldiers.
AI is simply another instrument, similar to guns, that will allow them to function more efficiently and handle advanced machinery.
Indeed, AI will transform the characteristics and possibly the caliber of jobs. However, it won't eliminate jobs or render them obsolete. While some types of roles may become unnecessary, new opportunities will also emerge.
I'm of the opinion that AI will lead to an increase in wages as it necessitates the acquisition of advanced and novel skills. Individuals will be compelled to enhance their own abilities or risk becoming outdated. Consider the evolution of the IT industry. Two decades ago, it was common for me to inquire if job applicants were familiar with SAP sizing and exchange sizing during interviews. Nowadays, such questions are no longer pertinent as they've become obsolete.
When individuals claim to be highly skilled in PowerPoint in their resumes, but are unfamiliar with integrating PowerPoint with AI, it indicates a lack of crucial skills. AI is a tool that will assist all of us in evolving. Ultimately, it's about staying pertinent, maintaining employability, and aligning with industry needs. To achieve this, personal investment is necessary.
Search for us on YouTube
Headliner Programs
Associated Reports
AI hallucinations can be resolved, and broad AI is expected in the next 5 years, says Jensen Huang of NVIDIA
Apple at last introduces MM1, its AI model for generating text and images
Microsoft recruits Mustafa Suleyman, DeepMind's cofounder, to head its new consumer AI division
Samsung and Rebellions, South Korean chip makers, aim to outshine NVIDIA
AI hallucinations can be resolved, and broad AI is expected in the next 5 years, says Jensen Huang of NVIDIA
Apple at last introduces MM1, its AI model for generating text and images
Microsoft recruits Mustafa Suleyman, DeepMind's cofounder, to head its new consumer AI division
Samsung and Rebellions, South Korean chip makers, aim to outshine NVIDIA
Available on YouTube
All rights reserved by Firstpost, copyright © 2024.
+ There are no comments
Add yours