Blog The Human Role in AI Development: Finding the Balance
By Carmen Taglienti / 15 Aug 2024 / Topics: Change and training Data and AI Modern workplace Digital transformation
In another AI in Business Podcast by Emerj episode, Chief Data Officer Carm Taglienti unpacks building trust in AI systems, AI workflows, responsible AI and more. In this blog, explore a summarization of the key points he covered or stream the podcast on demand.
The transcript below is edited and condensed for clarity.
What do terms like “responsibility” mean to you when it comes to AI at work?
There's a lot going on as it relates to responsible and ethical AI development. Humans are using these systems to help us to become better human beings — to learn more, to advance our own intelligence and logic — and I believe responsible AI involves using AI as a copilot. In that same vein, we shouldn't fall into the trap thinking AI is our servant and that we can sit back and let AI do everything for us.
The other side is the ethical side. I tend to bind the terms “responsible” and “ethical” together, but ethical AI use is being thoughtful and considerate about appropriate use cases. For example, using AI to get information about someone that may be a violation of PII, or using that information to steal someone’s copyrighted material — are use cases with ethical implications.
How can we build trust in AI systems?
I like the concept of “trust but verify” where we embrace AI to assist with human processes and remove the mundane — but we also allow for human-level reasoning decisions. An AI workflow can be built to include human assistance. Creating clarity around when humans get involved and establishing that we have a place in an AI work world goes a long way in building trust and working together with AI, not separate from or against it.
I often hear this concept of being the CEO of your own company in the realm of large language models. I’m guiding my “agents” to do whatever I need them to do as it relates to what I need to accomplish; they tell me when they’ve accomplished the task, and then I’ll take a look and verify that what they’ve done is correct and in line with what I want to do.
How do you think about measuring productivity gains from AI adoption?
AI is moving in a direction where we can answer questions at the speed of thought. When you find individuals that are really good critical thinkers, they want to keep learning more — so you can measure productivity simply by listening to people talk about the things that they were able to accomplish through technology.
We’re still in the early days, so how I think about productivity is through engagement. But in the literal sense, it’s as simple as answering the question: Were you able to get the job done in the time period which you thought you could get it done?
What will it look like when enterprises are effectively integrating human and expert feedback in AI systems — at the foundational level?
One of the biggest challenges for humans, and being an educator myself, I know that it takes a long time to accomplish what's called transfer learning. Basically, you go to class, you learn about cloud computing or cybersecurity, and it takes a long time to be able to do it. But if you have an expert that knows it, now you can take that transfer learning and move faster. In a utopian world, AI lends itself to a truly collaborative learning process where critical thinking is even baked into the model.
Hopefully, AI will eventually allow our brains to develop in a more reasoning and critical-thinking way because we won’t have to go back to ground zero every time when we’re learning something — we can leverage these intelligent agents to help us.
Discover thoughtful AI integration. Insight’s core capabilities span everything from modern data platforms to generative AI, advanced data analytics, governance and beyond.
Carm Taglienti
Chief Data Officer & Data/AI Portfolio Director and Distinguished Technologist, Insight
Carmen has more than 25 years of experience as a cloud computing, data science, data analytics and data management expert. As a principal cloud and AI architect for Insight, he focuses on cloud-based business solutions across multiple industries and technical domains. He also serves as an adjunct professor at Northeastern University Khoury College of Computer Sciences and the College of Professional Studies teaching graduate courses in data science, cloud computing and analytics. Carmen is a published author and speaks at industry and technology events.
