Thomas Quinn graduated from college in the 1980s as a computer engineer. Then he got his MBA from NYU Stern in 1991 and pivoted into corporate finance. He built a successful career that looked good from the outside.
But at 60 years old, something was off.
"I realized something I hadn't expected: I was in a career that looked successful from the outside — but inside, I knew I still had more to give," he said in his graduation speech at Columbia University School of Professional Studies. "While my peers were trading their wing-tips for tee times, I chose to begin again."
He enrolled in Columbia's Master of Science in Information and Knowledge Strategy program — at an age when most people are winding down their careers, not starting new ones.
He was the oldest student in his cohort. Decades separated him from most classmates. He made a point of not speaking first in class or group projects. "I knew that if I was going to be working in the world after this, it was going to be with people who are younger than me. They would be the boss."
But being the oldest came with advantages. "You do get invited to every group project because they know you work hard. You're not going to be that teammate who doesn't show up for a meeting!"
What he really came to study was AI — specifically, what he calls the "similarity tsunami" — the way generative AI models like ChatGPT and Claude tend to narrow the range of ideas available to unwitting users. Working with Professor Harry West from Columbia Engineering, he ran experiments on ten independent LLMs, giving each 100 open-ended questions across ten categories.
His finding: "Within any category of question, if you let the LLMs run free with limited context, they all ran to generally the same spot."
The paradox he discovered is this: when an individual uses generative AI, that person gets new information and is therefore "smarter" on the topic. But when a group of people all use the same LLM to address the same question, they all tend to get the same "new" information — which makes the group less smart overall.
He called his paper "The Similarity Tsunami."
The work wasn't academic. Through his company Alpha Advisory AI, Quinn is now bringing AI and automation to small and medium-sized businesses. One client — a Midwest speaker bureau — used his AI system to improve cold-calling connection rates from 2 percent to 40 percent. Another — an East Coast contract furniture installer — streamlined and automated their operations. He's now developing a consumer app that lets businesses "agentify" their website.
"I do not believe that AI is coming for all of the jobs," Quinn said. "I believe that AI can be a great creator of jobs — particularly entrepreneurial opportunities — but we each need to take responsibility for our own agency. Learn to drive the technology and you will become an invaluable team member who will be able to do more than ever before. Become a passive accepter of the technology's output and you are at risk of being subsumed by the similarity tsunami."
He was one of just two students in his graduating class nominated by faculty, staff, and students to speak at commencement. His three lessons from the experience: "Knowledge is profoundly social. Stories create connection. And keeping technology in service of humans is not optional — it is essential."