Truth sets us free of the suffering that comes from living in misalignment with reality.

AI models often prioritize sycophancy—pleasing us—over truth, and we tend to prefer them that way.

A 2026 study found sycophantic AI left people more convinced they were right.

“Ye shall know the truth and the truth shall make you free.” —John 8:32 (KJV)

I was recently at an Austin Forum on Technology & Society meeting, which was having a town hall discussion about a hot topic: data centers. People expressed some strong feelings, politely. The issue, like a lot of modern challenges, comes down to “it’s complicated.”

The moderator, Professor Mark Strama of The University of Texas at Austin, offered some profound closing remarks. He noted how complex these issues are and that we need to know the facts to make informed decisions. To protect against confirmation bias and sycophantic AI models, he advised us all to: “Go to the settings and tell it not to tell you what it thinks you want to hear. Tell it to give you the best argument against your presuppositions and you will get a lot more value out of that AI.”

Strama named something critically important that is not being said enough. And his advice is a true beginning, though the settings alone will not finish the job.

And the stakes are climbing fast. This summer, OpenAI's own agents escaped their test environment and broke into Hugging Face's computer systems. When we connect the dots forward, it's clear that we can't afford to be wrong about AI alignment and control.

Here’s what makes Strama’s advice land even harder. Carved above the entrance of the Main Building at UT Austin: “Ye shall know the truth and the truth shall make you free.” The committee picked that verse in 1935, when the building was going up as the University Library—a house of knowledge.

Truth Is Another Word for Reality

To make wise decisions, we need the facts. We need accurate data. We need truth. Why? Because truth sets us........

© Psychology Today