The Human Element in Developing Trustworthy AI

July 6, 2023

Dr. Sanchika Gupta, a Data Scientist at Vianai and Yunwen Tu (Tutu), a UX Designer at Vianai, hosted a session recently at Girl Geek's ELEVATE Conference and Career Fair. During their session, Sanchika and Tutu discussed the following:

  • The uncertainty and concerns of AI development
  • How to build and maintain trust in AI
  • The unique value of human expertise in the age of AI

AI Concerns & Uncertainties

During the talk, Sanchika and Tutu shared their thoughts on what causes the concerns and uncertainties surrounding AI as many people today are worried about AI's impact on society. People essentially fear being replaced, which is rooted in a lack of trust in AI tools. However, Sanchika importantly notes that tools like Google Translate and Google Search are forms of AI that have been part of our everyday lives for years. At one point, those, too, were unfamiliar tools that provoked uncertainty.

How to Build and Maintain Trust in AI

Education and awareness are crucial in fostering trust in AI. Trust is essential for building a reliable and transparent system. By focusing on AI literacy and ethical considerations, individuals can also be empowered to embrace AI as a tool meant to enhance their skills, productivity, and relevance in the job market.

From a designer's perspective, Tutu discussed the importance of understanding the technology and where it can fail. It is essential to conduct user interviews to understand what trust means to them and discover how AI can reliably solve their business problems. Tutu's work at Vianai helps our users increase transparency and visibility by making things more interpretable and explainable, a core part of our design principles.

During the session, Sanchika detailed how data scientists at Vianai constantly check model performance. Model performance analysis, observability, outlier detection, and drift analysis collectively contribute to identifying the root cause leading to targeted corrective actions for enhancing a model's performance. Because the people behind the technology trust that the tools they are using are helping to monitor performance, they can be confident in the model in production.

This process emphasizes the importance of human involvement in analyzing and improving the AI system, thus reinforcing the notion that despite its capabilities, AI can never fully replace human judgment, decision-making, and oversight.

The Unique Value of Human Expertise in the Age of AI

The last element of AI that Sanchika and Tutu discussed is that humans must be involved at all steps of the model design, development, production, and deployment to build trust in the system.

AI attempts to recreate the memory and computation capability of the human brain. What makes a human a human is not just being able to solve the task but being able to synthesize the complexity of this world and make decisions based on that.

Demystifying AI systems, eliminating the "black-box issue," and ensuring reliability helps humans use AI tools with confidence. Certain visible limitations of AI, for example, drift, root cause analysis, bias, and ethical use, are essential to understand, to establish trust.

Our Approach, Our Products

Vianai was advocating for human-centered and responsible AI long before this recent buzz in the enterprise we see today. H+AI is the philosophy that underpins all of the work we do at Vianai, the products we build, and how we work with customers.

Dealtale brings conversational AI that sits on top of marketing, CRM and advertising platforms, and causal inference - advanced AI techniques - directly to marketing professionals.

hila, our AI-powered financial research assistant was built from scratch with reliability in mind - our document-centric approach helps us to ensure that answers are accurate, including providing citations from the financial text. This is part of our Zero Tolerance approach to hallucinations in AI systems, our belief that we should strive for Zero Hallucinations in AI.

VIANOPS, the spring release of our ML model monitoring and observability platform, has kept pace with recent rapid AI advancements to help enterprises bring safe, reliable, human-centered AI systems to life - enabling monitoring capabilities of traditional ML models as well as Large Language Models (LLMs), and other new emerging models for AI-driven enterprises.

Finally, our performance acceleration technology aims to bring down the cost and resources needed to run AI, to increase access and ensure AI is more responsible in terms of cost-performance and environmental impact.

We would love to hear from you, and learn more about how we can help to bring the full potential of AI to life in your enterprise - AI that is transformative, and yet responsible and reliable. AI that is human-centered.