Everyone Sounds Smart Now, and I Don’t Know Who to Trust
Manoj Palasamudram
SubscribeSign in
Everyone Sounds Smart Now, and I Don’t Know Who to Trust<br>I noticed something recently, and I can’t unsee it.
Manoj Palasamudram<br>Jun 11, 2026
Share
When a colleague gives a sharp answer in a meeting, when a candidate handles a tough question, when a leader makes a confident call, I used to think, this person is smart. Now my first thought is, I wonder what’s running in the background.<br>Maybe you’ve felt this too. This essay is me trying to put words to it. I’m not saying AI is bad. I’m a software engineer, I literally build AI tools for a living. What I’m saying is something more uncomfortable. AI is making everyone seem intelligent, and I think that is breaking how we figure out who actually is.<br>What I observed
A few years ago the worry was simple. AI writes your emails, so good writing no longer proves a good mind. That already feels old.<br>Look at what I see around me now. AI joins meetings and summarizes them before the coffee is done. It listens to sales calls and feeds people the perfect reply, live. It preps talking points, guesses the questions, drafts the decision, rehearses you for the debate. There are tools that whisper answers during interviews, invisible to screen recording. Some can even speak in your voice.<br>I see it closest in my own work. Pull requests show up clean and well commented, and then in review the author can’t explain why a line exists. The code looks like understanding. It isn’t always.<br>So it’s not helping with writing anymore. It’s helping with the whole person. The conversation, the presence, the judgment, in real time.<br>And here is the thing I keep coming back to. We never had a direct way to check intelligence. You can’t open someone’s head and look inside. We always judged minds by performance. How clearly someone explains. How they handle a surprise question. How fast they connect ideas in a room. We trusted those performances for one reason: they were hard to fake. The effort was the proof.<br>Every one of those performances can now be assisted. Quietly. Live.<br>Sounding intelligent used to be evidence of being intelligent. Now it’s a monthly subscription. Am I wrong about this? I really might be. But I don’t think so.<br>The part that bothers me most
You’d think honesty would fix this. From what I’ve read, it doesn’t. It backfires.<br>Studies keep finding the same pattern. People can’t detect AI involvement, we guess about as well as a coin flip. But the moment someone admits using AI, trust drops. So hidden use gets rewarded, and honesty gets punished. We’ve quietly built a system where the smart move is to use AI everywhere and mention it nowhere.<br>Which means the people who look most brilliant right now? Some really are. Some are just the best at hiding the wire. I can’t tell them apart, and I don’t think you can either.<br>Why I think this matters
If this were only about emails sounding nicer, fine, who cares. But think about where we trust people based on how intelligent they seem. Two places come to my mind: leadership and politics. In both, we give someone power because their performance convinced us their judgment is good.<br>Companies have already noticed. They’re flying candidates in for in-person interviews again, at real cost, because they’ve quietly accepted that nothing said over a screen can be trusted. In tech interviews it’s already an arms race. Tools feed candidates answers on screen, and interviewers respond by asking people to close their eyes and talk, or to walk through their own code line by line. One recruiter described how she catches the assisted ones: “I’ll hear a pause, then ‘Hmm,’ and all of a sudden, it’s the perfect answer.”<br>That pause says everything. The gap between the performance and the person.<br>Politics shows where this can go. An official US government health report was caught citing studies that don’t exist. Written with AI, fingerprints and all. A confident, official-looking document with nothing real under parts of it. Now picture the same thing, but it’s a person. A candidate whose every public moment is assisted, whose unassisted judgment nobody has ever seen. We’d be electing the prompt.<br>The counterargument I keep hearing
People tell me leaders were always helped. Speechwriters, advisors, briefing books. No leader was ever a solo act. Fair point, and like I said, I’m not against the tools.<br>But the old help was visible and slow. The speechwriter was a known job. The advisors sat in the room. And most importantly, there were always unassisted moments. The live debate. The crisis call. The question nobody prepared for. That’s where the real judgment showed.<br>What feels new to me is that the help is now invisible, instant, and reaching into exactly those moments. The unassisted moments are dying out. And some early research hints at a second problem: the more we hand our thinking to AI, the weaker our own thinking gets. So the gap between seeming smart...