THU JUL 20 2023

PD Singh: Key Takeaways On Navigating AI Frontiers

by Bithika Bishesh

As a part of our Decoders Series, we had the privilege of speaking with PD Singh, Founder of a new Stealth AI startup, Former head of AI products at UiPath and Microsoft. Singh, with his 20+ years of significant experience in the realm of AI, has sculpted a track record of pioneering innovative AI solutions and leading them to a global scale.

In this first installment of a two-part series, we delve deep into the essential takeaways from our discussion with Singh on navigating the AI landscape. 

Takeaway 1: Embrace the Dawn of Domain-Specific Artificial General Intelligence

As we trace back the evolution of AI, it's clear we've come a long way. From the days of logic programming to the advent of Bayesian learning and the semantic web, AI has constantly evolved, setting the stage for groundbreaking advancements.

The deep learning revolution in 2008 marked a significant turning point, showcasing the immense potential of AI to address real-world use cases. 

This technological leap led to a fascinating debate about the distinction between AI and Machine Learning (ML). In the case of ML, there's data input and a desired output, with a process, be it a simple Bayesian model, Logistic regression, or a complex Neural Network that maps the input to output. However, despite the complexity of the process, this was still ML. This technology could not independently learn anything new unless subjected to a retraining loop, which was an intricate and time-consuming process.

Conversely, AI is defined as a system that starts to learn independently, either through online or real-time learning. We are at the threshold of witnessing systems demonstrating this level of autonomy, often coined as Artificial General Intelligence (AGI). While expecting a single AGI system to perform all tasks may be far-fetched, the emergence of domain-specific AGI is a foreseeable future, promising to usher in unprecedented changes. Businesses need to prepare to leverage this impending transformation in AI.


Takeaway 2: Join the Open-Source Revolution in AI

The AI research field is bustling with transformative open-source initiatives. Open-source projects are not just crucial for AI innovation; they're democratizing it. Developers and startups should embrace these open-source AI resources to boost their innovation efforts and stay at the forefront of the AI movement.

To fully capitalize on the advancements in the open-source domain, it's crucial to monitor the diverse range of models being developed. One area of particular interest lies in the exploration of foundational and large-scale models. With the introduction of new models by organizations like OpenAI or DeepMind, the open-source community swiftly responds with equivalent contributions, creating an inclusive and dynamic ecosystem.

This is the essence of the open-source model: it's fundamentally open and inclusive. By accessing open-source repositories, one can leverage pre-trained models, eliminating the need to train these models from scratch—a process that can be resource-intensive. 

Open-source resources are proving pivotal in accelerating the AI movement, fostering an environment of cooperation and inclusivity. This approach effectively breaks down barriers, allowing advancements to be shared freely rather than monopolized within individual companies or domains, embodying the true spirit of open-source innovation.

Takeaway 3: Aim to Achieve the Perfect Balance in AI Integration for Sales

Careful calibration of AI systems in alignment with a company's goals and values is paramount. Sales teams must strategically balance customer satisfaction with crucial business objectives when integrating AI into their operations.

In an enterprise setting, the key to successful AI implementation lies in customizing the objective function to align with the specific business outcomes. If a business is customer-centric, customer satisfaction should be central to the AI models' objective functions. This aligns the behavior of AI with the values and goals of the company.

However, different businesses may have different objectives. For instance, if a business aims to meet aggressive growth or revenue targets, these factors must be integral to the objective function. Overlooking this could lead the AI system to prioritize user happiness over these critical business objectives, potentially undermining the company's goals. Achieving the right balance is crucial to ensure that AI serves the needs of both the customer and the business effectively.

Conclusion

Navigating the intricate and rapidly evolving landscape of AI requires a deep understanding, flexibility, and a vision for the future. As PD Singh shared from his vast experience in the field, we are on the brink of significant advancements in AI that could lead us to domain-specific Artificial General Intelligence. The role of open-source projects in this journey is pivotal, fueling innovation and inclusivity in AI research.

However, the incorporation of AI into sectors like sales doesn't come without its unique set of challenges. Balancing customer satisfaction with unique business objectives becomes critical in these scenarios, ensuring the full potential of AI is harnessed without undermining essential business goals.

Aviso's AI exemplifies this balance, leveraging Generative AI to enhance sales processes and communication while addressing the challenges posed by GPT and other large language models (LLMs) models

To know more, book a demo with Aviso now.