DL Opinion columns

Human oversight of AI 2026 – by Deborah Levine

 originally published in The Chattanooga Times Free Press

It was a true gift to have relatives visit for the New Year. My step-daughter cooked for us non-stop and her hubby, an IT professional, taught me more than I ever imagined about websites. And he teased me with all kinds of terms, phrases, and acronyms knowing that I’d be clueless. But the teasing stopped when we talked about Artificial Intelligence (AI) and what it means for our future. 

Yes, I’ve worked for a company that trains AI, but I can still be intimidated by the topic.  Increasingly, I agree with my neighbor across the street who says, “I use AI for my work, but I feel it’s over my head.” That’s how I felt when The Times Free Press arrived the next day with the article “Meet the AI agents of 2026 – Ambitious, overhyped and still in training”. First, I’d never heard the term “AI agent”. The article’s definition took time to digest: “An AI agent is a software system designed to plan and execute tasks autonomously; make decisions and interact with digital tools or environments with minimal human oversight in pursuit of a defined goal.” Sounds amazing! No wonder that Meta just bought AI agent Manus for $2 billion. The marketing of AI agents is going through the roof with claims of automation and cutting costs, especially getting rid of the burden of expensive employees.  But maybe it’s not time yet to worry that AI agents will replace human beings and lots of jobs will disappear. The article underscored that AI agents are overhyped and should be used as tools and not replace human since “the gap between promise and performance is already becoming visible.” The writers wondered: “…not whether AI agents will be regulated, but whether oversight will arrive before harm becomes normalized.” And much of that harm comes from the not just the lack of human input, but the lack of human skills.

Not surprisingly, my inbox is full of invitation to webinars, conferences, and seminars teaching AI skills. There was an online course, Strategic AI Leadership in Business, designed to provide leaders with the frameworks, fluency, and structures needed to oversee AI adoption. I was also invited to an in-person university course in India that builds strategic clarity/integration and organizational readiness. It was impressive that the course also included, “governance, responsibility, accountability, and, most intriguing, ethics. 

I suspect that the ethics and values governing AI will be a major topic for 2026. I just got a LinkedIn alert about a Huffington newsletter featuring this article: “The Human-AI Alignment Problem”. Huffington quotes Sam Altman, the co-founder and CEO of OpenAI, who launched ChatGPT in 2022. Altman claims that ChatGPT is trained on the “collective experience, knowledge, learnings of humanity.” Sounds impressive, but what does that means?

Good question…and Huffington asked about the meaning, focusing on how Altman’s description of AI training implies the inclusion of human values. She states that our human connection to the spiritual foundation and values of both Western and Eastern civilizations is iffy these days. Can AI be trained to be ethical?

I turned to a colleague, Dr. Carlos Cortés, who pointed out that the “collective experience, knowledge, learnings of humanity” is largely based on English language sources. He reports that about 97% of the world’s languages are considered to be “low resource” and don’t get much attention. In such a slanted knowledge arena, being ethical is unlikely without human supervision across cultures and languages. 

Bottom line 2026: AI requires human oversight so let’s pay attention and learn more about AI. And if you see something unethical… definitely say something. 

Editor-in-Chief
Latest posts by Editor-in-Chief (see all)