We're finally entering a new era of AI. Confidence. Not confidence in the models, but confidence in ourselves. Six months ago, if you remember, people were asking, can AI do this? Now we're all asking, what else can it do for me? That is a huge shift. The technology is improving at lightspeed, but people's willingness to experiment is improving even faster. I have a question. What's the last thing that you asked AI to do to take off your plate?
Let me sit on this, because the shift I'm describing is easy to miss precisely because it happened inside people, not inside the technology, and the things that happen inside people never get a press release.
What actually changed in the last six months?
The question changed. That's the whole observation, and it's bigger than it sounds.
"Can AI do this?" is a spectator's question. It holds the technology at arm's length, puts the burden of proof on the machine, and keeps the asker safely in the audience. It's the question you ask about something you don't yet trust, and, more honestly, about something you don't yet trust yourself to use. Every new technology lives in that question for a while. Most people asked it about AI for years.
"What else can it do for me?" is an operator's question. The burden of proof is gone. The asker has already seen it work, already handed something over, already gotten an afternoon back. Now they're hunting. They're walking through their own day looking for the next thing to hand off. Same technology, completely different person standing in front of it.
And here's why I call this era confidence in ourselves rather than confidence in the models. The models were capable of most of this a while ago. Capability was never the bottleneck for long. Permission was. What people lacked wasn't a better tool. It was the internal go-ahead to try, to look a little foolish, to hand real work to a machine and own the result. That go-ahead is what's spreading now, person to person, faster than any release cycle.
Why is willingness moving faster than the technology?
Because confidence compounds socially and technology doesn't.
A model improves on whatever schedule its makers ship. But confidence spreads every time one person watches another person hand something off and get their time back. The colleague who stopped writing first drafts by hand. The friend who let AI plan the trip. The agent down the hall who left the office at four because the paperwork didn't need her anymore. Every one of those moments recruits the next experimenter, and every new experimenter creates more moments. That's a curve no lab controls, and right now it's the steeper one.
I see it in my own industry every week. Real estate agents who a year ago wouldn't touch AI with someone else's phone are now the ones asking me what else it can take. Not because a benchmark moved. Because someone they trusted showed them what their own week could feel like, and the question flipped from can it to what else.
What should you do with this?
Notice which question you're asking, because it tells you exactly where you are.
If you're still asking "can AI do this," the honest answer to almost any version of that question is now yes, and the more useful move is to stop asking it in the abstract and hand over one real task today. Small. Low stakes. Yours. The first handoff is where the question flips, and it never flips from reading about it.
If you've already flipped, then keep pulling the thread. The people getting the most out of this era run a standing audit on their own plate: what's still on here that doesn't need me? Every week the answer changes, because the technology moved again, and the only way to know is to keep asking.
The technology is improving at lightspeed. Your willingness is the thing that decides whether any of it reaches your actual life. That's the era we're in now, and it belongs to the people asking the second question.
So, my question to you. What's the last thing you asked AI to take off your plate?
*Judd Hoffman is CEO and Co-Founder of Ethica AI, building AI-powered tools for real estate transaction workflows.*