Scaling Generative AI in the Enterprise – A Fireside Conversation with Bratin Saha

Moderated by Iavor Bojinov and Edward McFowland III

Since the launch of ChatGPT, GenAI has taken center stage, prompting business leaders to explore its potential for creating value. In a recent stimulating discussion featuring Bratin Saha, Iavor Bojinov, and Edward McFowland III, the conversation delved deep into the radical potential of GenAI and its seismic impact on businesses.

Business Horizons Illuminated by GenAI Models

Bratin highlighted several areas where GenAI creates unique opportunities in enterprises:

  • Automating Repetitive Tasks
  • Enhancing Productivity and Fostering Creativity
  • Innovative Business Models
  • Reducing the Digital Divide
  • Optimizing for Specific Domains

“The capabilities of machine learning models have surged by a staggering 1600 times, an unprecedented rate of advancement. These models are now trained on trillions of data points, whereas an average human, throughout their lifetime, might be exposed to a mere billion or two billion words.”

Saha addressed the core aim is to transition from a bespoke, manual approach to a standardized, automated, reliable, and robust system – essentially industrializing the entire process. The choice between generative models and specialized models will be primarily driven by the use case, the associated cost considerations, and the potential impact. While there are use cases where generative models are indispensable, there are also instances where smaller specialized models are more efficient.

How to arrive at these smaller models? Saha explained two main ways. You can either start with a smaller model and train it to suit your specific needs, or you can begin with a larger foundation model and employ techniques like quantization and compression to reduce it to a more compact size. The approach you choose will depend on the particulars of your use case.

GenAI’s Intricate Challenges: The Transformative Impact on the Workforce

“Generative AI is not merely replacing human work but rather augmenting it with human oversight.”

Saha provided illuminating insights into the myriad business opportunities that GenAI models present. The foundation of the discussion on how this technology is penetrating various industries stems from the shifts in the creation process in people’s work.

“This creative process typically comprises two main elements: first, you have thoughts and ideas in your mind, whether you’re writing a paper, creating software, crafting a painting, or penning a book. These thoughts reside within your mind. The second aspect involves taking those thoughts and translating them into tangible output.” Which part of this creative process is being replaced or augmented by GenAI tools?

As the conversation unfolded, Saha candidly addressed the multifaceted challenges presented by GenAI. Among the key quandaries discussed was the friction between aggregated creation and individual innovation. Drawing from a real-world example – research conducted by HBS in collaboration with BCG – the reality is that GenAI end-users often yielded outputs that displayed notable similarities. This pointedly underlined the importance of fostering diversity in ideas when harnessing the power of these transformative tools. The onus, therefore, lies on organizations to strike the delicate balance between the sheer potential of these tools and the preservation of the authenticity and uniqueness of individual creativity.

Augmentation through automation of mundane and time-consuming tasks could usher in a new era of efficiency and productivity in workforce. Additionally, the remarkable prospect of conversing with machines in one’s native language, notably in the realm of coding, presents an enticing shift in the employment landscape. Roles are set to evolve, demanding new skill sets while potentially automating significant work aspects.

Unraveling the Conundrum of Creativity and Ownership

The engaging conversation further unraveled the intriguing dimensions of creativity and ownership, particularly in fields like writing and acting concerning the attribution of content generated by GenAI. One of the angles to address this concern is policy considerations. Moreover, Bratin posed another question. “When we train a model and it generates output, there’s also the human element of reading papers, viewing artwork, and acquiring knowledge from various sources. This constant process of learning from external works raises intriguing questions. For instance, has my own writing style been subtly influenced by the countless novels and newspaper articles I’ve read over the years?”

A Vision for the Future of Business with GenAI

“If your company were founded today, in the era of GenAI, how would it differ from your current organization, which was established before the GenAI era?”

GenAI is far from being a mere productivity booster; rather, it stands poised to herald the next wave of innovation. Waiting for GenAI to mature could translate into missed opportunities and heightened vulnerability to disruptive forces. In this light, Saha challenged corporate leaders to re-imagine their organizations as if they were founded in the GenAI era, urging them to identify how they might differ from their current incarnations. This articulates the urgency for CEOs and board members to seamlessly weave GenAI into their strategic blueprints, ensuring that they stay agile and competitive in the ever-evolving corporate milieu.

Engage With Us

Join Our Community

Ready to dive deeper with the Digital Data Design Institute at Harvard? Subscribe to our newsletter, contribute to the conversation and begin to invent the future for yourself, your business and society as a whole.