Generative AI: Into A New World
One of the earliest direct mentions of the wheel comes from the Rigveda, believed to have been written around 1700 BC. There are earlier, more oblique references as well. For instance, in the ancient Sumerian poem "The Curse of Agade” from around 4000 years ago, the goddess Nisaba threatens to turn Agade into a place where chariots won’t turn.
Regardless of the exact year of its invention, the wheel profoundly changed the course of human civilization by allowing us to conquer distance and time. We were able to reclaim energies lost on travel and logistics and divert them elsewhe
re. Who knows how many other inventions were made possible because their progenitors didn’t have to break their backs carrying rocks, and were able to finish their day’s work faster.
Today, Generative AI offers us a similarly significant chance to move past the mundane and accelerate our civilization’s progress, much as the wheel did for our ancestors thousands of years ago.
AI has been around for longer than most people would believe. The term Artificial Intelligence was coined at the Dartmouth Conference of 1956, more than a decade before we’d set foot on the moon. However, the excitement generated by the field between 2015 and today, exponentially exceeds anything the field has seen in the previous 60-years. Companies continue to pour billions of dollars into research and development, while simultaneously integrating Generative AI within their processes, products, and DNA.
What is Generative AI, how is it different from the regular AI, and what makes it so special?
What Makes Generative AI So Exciting?
A simplified difference between AI and Generative AI, but one that will give you an idea of the difference between the two is this: Generative AI possesses the ability to go beyond the data it has been fed. In humans, this is perhaps what’s known as the faculty of imagination.
Traditional AI models usually rely on supervised learning methods where they are fed labeled data that comes with Input/Output pairs. They are also taught to minimize a predefined objective function, which is basically a quantification of the difference between the AI model’s predictions and the actual truth in the training data. Generative AI models on the other hand, go through unsupervised learning where, in the absence of Input/Output pairs, they learn to uncover patterns and relationships in the data without the need for human-labeling.
Delving into the exact machinations of Generative AI models and their working is a topic for another day, but in short, Generative models are capable of mapping the input data and then producing synthetic data that aims to be indistinguishable from the inputs. I can talk about bell curves and probabilities, but let’s just say that Gen AI models understand the logic behind the input data and can use that to create more of it.
In essence, traditional AI models are narrow, deterministic and designed to excel at specific tasks. Models such as transformer, GAN, autoregression or variational autoencoders make Gen AI broad in its capabilities, versatile and flexible, enabling it to produce creative and innovative probabilistic outputs. To take a real-world example, think of traditional AI as a builder following pre-set rules, while Generative AI is like an architect pushing boundaries and creating completely new designs by putting together all of her knowledge, expertise and experience.
An Inside Look at Generative AI in Action
A couple of weeks ago, Elon Musk announced the launch of xAI to compete against OpenAI, Microsoft, Google and other industry leaders. Within days of the launch, xAI shared that it had raised $6 Bn in funding. In India, Gen AI startups have raised over $500 Mn 2023 alone. Many of these startups are helping enterprises solve their business challenges.
While a lot of conversation around Gen AI has been driven by ChatGPT, Midjourney and other tools that have caught the public’s imagination, it can be argued that enterprises will unlock the real potential of the tech. There is also a remarkable buzz, with almost every tech company worth their salt either bringing in a Gen AI line of products and services or integrating the tech with their existing products.
From Adobe using Gen AI to augment its creative suite, to NVIDIA’s pivot that has seen it blitz ahead of Apple and become a $3 Trillion company, Gen AI is increasingly being seen as an immediate force multiplier in every area of business, right from drug discovery to customer support.
A Co-Pilot to Keep You Focused on What’s Important
Some areas where Gen AI is starting to have a significant impact include labor productivity, customer experience, and research and development. Companies are already deploying Gen AI alongside their developers, designers and writers to increase their efficiency while reducing the margin of error.
Imagine a co-pilot (which incidentally, is what MIcrosoft has named its AI assistant) that sits right next to you and does the bulk of the work, leaving you to focus on more strategic aspects of your role, and validate and confirm everything the co-pilot has done before moving it to production.
According to a recent study jointly conducted by Russian and Indian universities, using Generative AI in business has led to a 35% increase in productivity, a 20% rise in job satisfaction, while reducing costs by 40%. Good for business, good for employees, and good for customers. While we have been let down by overhyped technologies such as the Metaverse before, it seems like we have discovered a win-win-win formula in Generative AI.
Similarly, transformer-based models such as ChatGPT are helping enterprises drive high-quality marketing and sales collateral customized to specific customers, solve particular problems, and promise to improve RoI. You may have already come across AI-powered chatbots or even voice assistants that can help solve most of your queries on call, and redirect you to the correct human in case they encounter something outside of their expertise. Gen AI is also being used to enable proactive research and strategizing at a leadership level. Operating inextricably with data, Gen AI is enabling leadership and management functions with the insights they need to make smarter decisions.
Growing Pains On the Way to Gen AI Maturity
If there were a formal script for human civilization, right about now, we would be at a cliffhanger. Gen AI’s power and versatility have the potential to make it ubiquitous in the next few years and completely change the way we live, work, and operate. I wouldn’t be surprised if IoT soon evolved to cover devices enabled not just with the internet, but also with Gen AI. The Gen AI transition has already started with technology manufacturers and HiTech companies integrating Gen AI in smartphones, chips, and other devices.
There are risks associated with Generative AI and they are significant: IP concerns, copyright or legal exposure (proprietary data being used to answer user prompts), misuse of information, sophisticated phishing methods that use deepfakes, and factual errors or bias caused due to AI hallucination where the models invent stories that don't really exist, to cater to the user. There have also been a lot of concerns around job displacement, and how Gen AI will affect human/intellectual capital.
Back to the Future: An Era of Thinkers
My two cents on this topic are these: Rather than disruptive, Gen AI is going to be iterative. Following the footsteps of the internet (which started off as a way to send mails) we are going to discover viable uses for Gen AI and double down on them. There will be change in human capital, but the picture won’t be as bleak as many would have you believe. The nature of jobs will transform, and many roles that exist now won’t in the future. This, though, is a fact that has historically been true (In the Victorian era, there existed a professional class called the Knocker Uppers who went around waking people up in the mornings) and new roles will come up faster than the old ones disappear.
I strongly believe that with Gen AI making the “doing” part easier (You can now create websites, apps, and entire movies through prompts), philosophers and thinkers who are able to come up with original ideas, will thrive. This change in working models and habits has already started, with Gen AI forming an indispensable part of the work routine for most professionals, whether for creating briefs, for drafting communication, or for research.
Going back to the wheel analogy we used at the beginning of this article, the initial wheels were probably little more than a circular piece of wood. In fact, anthropologists are confident that wheels were used in pottery before logistics and the very first wheels used for motion, were used to move little more than a child’s toy.
If wheels can evolve, so can humans.
If you were paying attention last year, you’d have noticed how for a few weeks the conversation on business forums (and in business circles) revolved almost exclusively around Generative AI. This was around the launch of ChatGPT whose conversational capabilities gripped the public imagination.
The expectation perhaps was of massive, sweeping changes that would instantly redefine the way we work. The reality has been more sobering, and there are quite a few challenges that need to be overcome before Gen AI can really come into its own. However, the entire story has been interspersed with aha! moments as we keep discovering new avenues of value creation through Gen AI.
