The Cost of Switching: An AI Lock-In Experiment (Post 6 of 6)
So, Was the Juice Worth the Squeeze?
When I started this experiment, I thought switching would be a matter of dollars and curiosity. If ChatGPT was $20 a month and Claude or Gemini were also $20, then the cost of moving should be minimal. What I found instead was a completely different story.
The real costs weren’t financial. They were procedural (all the technical work just to move files), cognitive (the mental effort of relearning new tools), and trust-based (the fluency I had built with ChatGPT that takes time to establish).
Claude never even made it past the starting line. Its context window was too small for my exported history, no matter how many ways I sliced and reformatted it. Gemini worked better, thanks to its Google Drive integration but even there, the trust gap slowed me down. I didn’t have years of fluency built up, and every answer felt like a gamble. Right now I have feet in both camps and as they evolve over this fall it will be interesting to see how things pan out. But, I am an odd one. I carry an iPhone and Android, I have MacBook and CopilotPC, and don’t get me started on my browsers; Chrome, Safari, Edge, Dia.
Alongside that personal journey, the research gave me three clear theses to frame what I experienced:
The Trust-Based Lock-In Thesis - Once you’ve built reliability and fluency with a tool, switching feels like paying the “Trust Tax” all over again .
The Paradox of Choice Thesis - More options don’t make switching easier; they create decision fatigue that drives you back to “good enough” .
The Organizational Inertia Thesis - Companies that wait too long end up fighting grassroots lock-in. Employees have already chosen their tools, and now leadership has to ask them to switch .
So, was the juice worth the squeeze? For me, no. I ended up back with ChatGPT—not because it’s perfect, but because the switching costs were higher than the benefits I gained elsewhere.
That’s the paradox of AI adoption in 2025: the market is moving fast, tools are advancing, and yet the gravitational pull of trust and fluency keeps us anchored. Until portability, interoperability, and trust catch up, most people—and most organizations—will stay locked in.
If the AI you use every day isn’t perfect, what would it take for you to actually switch and is that change worth the squeeze?
