How to Future-Proof Your AI Projects and Implement Governance
Discover the impact of future AI Regulation with Sushant Hiray from our panel session ELC Annual 2023 — The largest conference for engineering leaders in San Francisco!
This is the second article in our 3-part series where we uncover the insights from our panel discussion at ELC Annual 2023. Let’s dive in!
Read the first article from our panel discussion with Grace Chu, former Senior Product Counsel at Adobe, on Responsible AI Development: Working Hand-in-Hand with Legal
In this blog post, we will uncover the following:
- Anticipating AI Regulation
- Navigating Compliance Challenges
- Strategic Readiness for AI Regulation
Let’s get started!
About Sushant Hiray
Sushant is the co-founder and chief technology officer at DeepAffects, a speech AI startup acquired by RingCentral. With a strong academic research background, Sushant has authored numerous patents and research papers, showcasing his expertise in the field. He currently leads the Machine Learning efforts at RingCentral.
How do you feel about the upcoming AI regulation?
Sushant Hiray begins by acknowledging the necessity of AI regulation as these systems become more prevalent. He emphasizes:
“Regulation is definitely needed. As we talk with data about people, data is being shared along. So we had a lot of data-related regulations; the next step is model relations. You have so many models in production already which are not regulated. So that’s the next thing.”
What internal discussions are happening now about complying with new regulations?
Sushant shares how he instilled a compliance mindset at RingCentral to prepare for compliance with new regulations. He helped establish an AI governance council, bringing together engineering, product, and legal leaders early to navigate evolving regulations.
“To ensure responsible AI practices, we have established an AI governance council. This council serves as a recurring platform for meetings, attended by myself, engineering leaders, product leaders, and notably, legal leaders within our organization”
How is RingCentral approaching model documentation? Do you plan to improve this in the future?
Addressing the challenge of model documentation, Sushant discusses the disparity between regulated sectors like banking, where documentation can run thousands of pages, and RingCentral’s more limited documentation. He highlights the importance of prioritizing data governance first, with model documentation as a subsequent phase as regulations expand.
Sushant suggests concentrating initial compliance efforts around data for organizations new to AI regulation. Robust data sourcing, consent, and management policies create critical foundations for broader compliance.
“I have taken the initial stride of documenting our data, including its lineage. Moving forward, the subsequent step entails focusing on the models itself “
Sushant urges early discussions around data practices, even if it feels premature. He emphasizes the importance of establishing discipline around data to prevent larger issues when regulations intensify.
Sushant’s best practices for organizations to ready themselves for AI regulation:
- Foster a compliance mindset with cross-functional governance bodies
- Increase model documentation progressively in line with emerging regulations
- Engage legal/compliance teams early when developing AI systems
- Plan for compliance needs to change over time rather than remain static
Final thoughts
As the AI landscape undergoes increased scrutiny, organizations deploying AI must strategically position themselves to operate within more regulated environments. By heeding Sushant Hiray’s advice on robust data governance, progressive model documentation, global regulatory monitoring, and cross-functional compliance collaboration, companies can effectively prepare for the new era of AI oversight.