Generative AI – The creative fuel for enterprise excellence

The evolution of Generative AI has been a captivating journey marked by notable developments and significant advancements in the field. It all began in the early days of artificial intelligence (AI) research, when scientists and researchers were just starting to explore the possibilities of creating machines capable of generating human-like content. The evolution of generative AI reflects the relentless pursuit of mimicking human creativity and intelligence through machines, with each era building on the accomplishments of the previous one. This innovative branch of AI opens up a universe where machines can reflect some level of human-like creativity, bringing us a step closer to the vision of truly intelligent systems.

OpenAI’s generative pre-trained transformer (GPT) models GPT-2 and GPT-3 (released in 2019, 2020) marked a huge leap in the field of Gen AI for text. They demonstrated the ability to generate coherent and contextually relevant human-like sentences, making them extremely useful for a broad range of applications from content writing to more intelligent conversational chatbots. OpenAI later launched in 2022 DALL-E, which is a deep learning model that can generate digital images from natural language prompts. Toward the end of 2022, OpenAI released ChatGPT, an intelligent conversational chatbot based on GPT, and the platform reached one million users within five days. It took Instagram 2.5 months, Spotify 5 months, and Facebook 10 months to reach the same milestone. Earlier in 2023, GPT-4, which is reportedly more accurate and has advanced reasoning capabilities, was launched. Premium ChatGPT users can now pay for access to the new GPT platform. Each of these recent milestones has brought Generative AI closer to its current capabilities, overcoming challenges related to computational capacity and power, data quality, and stability of training.

What is Generative AI?
Generative AI refers to AI technique that learns a representation of artifacts from data and models, and uses it to generate completely new and completely original artifacts that preserve a very close likeness to the original data or models, and these are built on foundation models. Large language model (LLM) is a type of foundation model that specifically focuses on natural language. LLM is trained on a very large amount of unlabeled data, using what is called a transformer algorithm. This training is augmented by a range of fine tuning (adapter) mechanisms that results in a model that can be adapted to a wide range of applications.

Key pillars for an enterprise’s Generative AI strategy
Link Generative AI enterprise goals with the organizational vision. Enterprises should clearly spell out their goals as it relates to Generative AI, and what benefits can be expected leveraging the power of this Technology. The next important step would be to clearly identify the success criteria and how it will be measured. This will help to fund the right use cases and these use cases can relate to goals around improving topline revenue growth, enhanced customer satisfaction, lower costs, increasing productivity of employees, improved service availability, etc.

Value capture and removing potential barriers than can hinder success. Having identified potential benefits to the enterprise in the vision stage, any strategic concerns that could hinder the enterprise ability to capture value in the way they have identified it should be flagged off. The projects that are aligned to the enterprise goals are most likely to succeed. One barrier to the success of a Gen AI project could be the issue around change management as it relates to an enterprise’s internal processes and also for people to have the required skills in order to work with Gen AI tools. Enterprises can either run skilling programs internally to up skill existing employees or the option to partner with niche Gen AI solution providers including Gen AI startups to work on Gen AI pilots that has the potential to scale.

Addressing risks. Any type of AI comes with a broad range of risks, including regulatory and reputational. Gen AI also carries new types of risks, such as hallucinations and biased and inaccurate results. Hence the Enterprise has to understand the continuously evolving regulatory landscape. There is a clear case to enable collaboration between the AI practitioners and legal, risk and security members to evaluate the specific use case feasibility and the acceptable risks. It would require the creation of an AI governance office, which serves an independent audit committee to review the results. It is also important to acknowledge the threats against AI, posed by both malicious and benign actors within the enterprise. The enterprise has to beef up security across enterprise security controls, data integrity, and AI model monitoring and, if required, to leverage external resources to help secure enterprise AI systems.

Applications of Generative AI
The applications of Generative AI now span a broad range of industries and fields. One of the key benefits of Generative AI is that use cases can be general purpose across all industries, and not restricted to a specific domain.

For example, in the marketing and sales function, Generative AI can be used to create the first draft of text documents, personalized marketing, and summarizing text documents. In the area of product and service development, Gen AI can help in Identifying trends in customer needs, drafting technical documents, creating new product designs, etc. In the area of service operations, it can help in the use of chatbots for customer service and also forecasting service trends or anomalies.

In the healthcare industry, Generative AI is used to create synthetic data for research, especially in the drug discovery lifecycle, allowing scientists to move healthcare forward while maintaining privacy regulations. In the financial and Fintech industry, Gen AI use cases would revolve around improved customer experiences, fraud detection and prevention, and enterprise risk management. In the entertainment industry, it is used to develop new video game levels or generate special effects for movies. In the information technology industry, technology service providers like Cognizant, Infosys, Wipro, etc., are embedding Generative AI into most part of its operations and ensuring software developers are embracing tools like ChatGPT to help them in code development, which includes code analysis, code migration, documentation and automated testing. ChatGPT is also fused into their internal knowledge management systems and service delivery platforms, which caters to their customer requirements across the world. Microsoft have fused AI into their Edge browser and Bing search engine. Microsoft also announced that it is incorporating AI, which is called Co-Pilot into its entire office suite – Word, Power Point, Excel, Outlook, and Teams. Adobe Firefly is a new set of Generative AI tools that can allow content creators to generate very good quality images and text effects.

The automotive sector exhibited high potential as the most lucrative segment amongst the manufacturing companies in the year 2022 and 2023, as per a report by 3AI, a bespoke AI consulting and research firm. Potential Generative AI applications include product design, process optimization, quality control, and predictive maintenance. Generative AI can also enable automotive manufacturers to enhance production efficiency, reduce costs, and improve overall product quality. For example, Volkswagen has employed Generative AI to enhance automotive design. Their AI algorithms analyze all the design parameters, manufacturing constraints, and safety regulations to generate optimized designs. This technology has enabled the development of innovative vehicle structures, improved fuel efficiency, and enhanced safety features. A second example is Adidas that utilizes Generative AI to design and customize athletic footwear. Their AI-driven platform, Strung, analyzes athlete data, performance metrics, and design preferences to generate unique and personalized shoe designs. This technology enables Adidas to create products that cater to individual needs and preferences, offering a personalized customer experience.

Amongst the Indian enterprises, automotive and farm equipment business of Mahindra and Mahindra has taken an early lead in deploying Generative AI in its operations. Through the complete automation of robot and heavy machinery maintenance on the factory shop floor, the company has reduced downtime.

Zomato, one of India’s largest restaurant aggregator and food delivery firm, is experimenting with Generative AI to enhance customer experience. The company is planning to hire engineers in ML, data science, and NLP roles to develop AI-based products. By integrating Generative AI into their existing operations, Zomato aims to handle increasing loads more efficiently and effectively in customer communications within the app. Zomato has already utilized neural networks for predicting food preparation and delivery time, while its subsidiary Blinkit has employed machine learning to optimize supply chain cost and reduce delivery times. Zomato is collaborating with companies like Google, Adobe, OpenAI, and Microsoft for AI product development.

The Generative AI opportunity
The Generative AI market is expected to grow to USD 1.3 trillion over next 10 years from a mere USD 40 billion in 2022, as per Bloomberg Intelligence (BI). 70 percent executives believe that Generative AI will enable enterprises to widen the scope of the roles of knowledge workers. Nearly all executives (96 percent) recognize that Generative AI is a key topic in the boardroom, with the majority confirming that there is very strong support from the top leadership, as per a report from Capgemini Research Institute. Generative AI is expected to generate an economic value worth USD 2.6–4.4 trillion annually, of which around 75 percent is expected to be concentrated in software engineering, customer operations, product and R&D, and sales and marketing, which are core service lines of many technology service providers in India, as per a report by NASSCOM in association with Mckinsey. As per the same report, emerging Gen AI use cases will enhance productivity and growth for tech services players with reimagined service offerings over next 5 years. Most of the large technology service providers in India have integrated Generative AI into their offerings and are working on a number of Generative AI pilots with their Enterprise customers. We are also seeing the emergence of a number of Generative AI startups, who are keen to exploit this massive opportunity. As per the NASSCOM Generative AI Landscape in India report, released in the middle of this year, there are 60+ Generative AI startups as of Q2 FY23 and funding to the tune of USD 475 million has already been committed. 52 percent of these startups are focused on text, images, and video and 70 percent of the startups are Gen AI native and are developing in house solutions.

Conclusion
Generative AI has the potential to be more transformative than all the previous waves of AI, and it will need the entire ecosystem, which is the startups, industry, investors, and the government to work together and a mission scale effort to build India’s niche in the global Generative AI landscape. The startups need to co-innovate with industry and the cloud providers to build quick scale, conduct thorough impact assessments to mitigate potential legal impact and most importantly build specialist talent pipeline through academic partnerships and internship opportunities. The Industry would need to create awareness drives to demystify Generative AI for consumers and businesses and champion and evangelize responsible AI framework, and also partner with industry associations to skill incoming talent at scale. The investor community should invest in concepts fulfilling major whitespaces in the Indian context, but with global market potential, and scrutinize for responsible AI implementation. Finally, the government should identify high-priority use cases across government departments, focusing on integration with public services and existing DPI platforms. There should be a huge focus on obtaining funding for local compute resources, Indian datasets, and innovation in the space of quantum and AI. Since the Digital Personal Data Protection Bill has been passed in Parliament and is now an Act, ensuring a timely enforcement over the next months will be extremely critical. The government should also make attempts to lobby in international fora for globally consistent responsible AI guidelines.

The future of Generative AI presents tremendous potential and is truly a disruptive technology. However, as with any emerging technology, there are challenges that need to be addressed. Ethical implications have to be carefully considered, such as the responsible use of AI-generated content and the threat of fraud through AI-generated voices. Intellectual property rights, transparency, and interpretability issues require utmost attention to ensure fair, responsible, and accountable use of Generative AI. As we navigate the future of Generative AI, we must balance harnessing its tremendous potential and at the same time address the associated challenges to create a responsible and beneficial AI-powered ecosystem.