I'll tell you the honest answer up front, and then spend the rest of this trying to convince you it's the wrong question.
The honest answer is: for a lot of things, yes, and it already has, and pretending otherwise is a comfort I can't really afford to sell you. For a lot of other things, no, not soon, and possibly not ever, and understanding exactly where that line falls is, I think, the single most useful thing a working person can do right now.
But the question itself, "can artificial intelligence replace real intelligence," smuggles in an assumption that quietly poisons the whole conversation. It assumes the two are the same kind of thing, competing for the same ground, so that one can simply move into the space the other vacates. I don't think that's what's happening. I think something stranger and more interesting is going on, and you can only see it if you stop asking who wins.
What the machines are genuinely good at
Let me not insult your intelligence by underselling theirs.
I use these systems most days. They draft, they summarize, they translate, they write code, they explain things patiently at two in the morning. They can hold more facts in view at once than I ever will, and they never get tired or bored or defensive. On a huge range of tasks that we used to treat as proof of human cleverness, they are now simply better, or fast enough and cheap enough that better stops mattering. If your sense of your own worth was built on being the person in the room who could produce a competent summary, or a clean first draft, or a tidy analysis of something fairly bounded, then yes, something you relied on has genuinely changed, and I'd be lying to tell you otherwise.
So this is not one of those reassuring essays that pats you on the head and says the robots will only ever do the boring bits. The robots are coming for some distinctly non-boring bits, and they're already doing them, and I think the people who feel a quiet unease about it are reading the situation more accurately than the people who don't.
But watch what happens at the edges
Here's where it gets interesting, and here's where the word "replace" starts to fall apart in your hands.
These systems are extraordinary at producing answers and strangely helpless at knowing whether the answer matters. They can tell you fifty things about a situation and cannot tell you which of the fifty is the one that will keep you up at night. They generate fluent, confident, plausible output, and they have no felt sense of when plausible is dangerous. Ask one to help you make a hard decision and it will lay out the considerations beautifully and then, at the exact moment a decision actually requires something, hand it back to you, because deciding means taking responsibility for being wrong, and responsibility is not a thing you can compute. It's a thing you carry.
I've come to think that's the real line, and it isn't where most people draw it. The line isn't between tasks that are hard and tasks that are easy. The machines are often best at the hard-looking ones. The line is between producing something and being accountable for it. Between the answer and the ownership of the answer. A model can generate; it cannot stand behind what it generated when the thing goes wrong, because there is no "it" there to stand, no career to risk, no reputation to lose, no 3am dread about the number it got wrong. We keep testing these systems on whether they can produce the output. We should be testing them on whether they can be the one who's responsible when the output is acted upon, and on that test they score a flat, permanent zero.
The thing we keep forgetting about ourselves
There's a second half to this that I rarely see said out loud, and it's the part that actually matters for how you should feel on a Monday morning.
The anxiety about AI replacing us is, underneath, usually an anxiety that we were only ever the sum of our outputs. That a person is a kind of machine for producing deliverables, and a better machine has arrived. But almost nobody actually believes that about the people they love. You don't value your closest friend for the quality of their summaries. You value them for judgement, for the particular way they see things, for showing up, for the fact that when they tell you something is a bad idea you trust the whole history of the person saying it. We know, in our actual lives, that intelligence was never really about output. We only forget it when we start talking about work, because work spent a century measuring us by output and we half-believed the measurement.
What AI is doing, if you look at it clearly, is stripping away the part of human work that was always machine-like, the part that was output for output's sake, and leaving exposed the part that was never machine-like at all. The judgement. The taste. The willingness to decide and be wrong. The sense of what's worth doing in the first place, which no system can generate because it has no stake in the doing. That part isn't being replaced. It's being concentrated. It's becoming, if anything, the whole job, where before it was maybe a third of it buried under the production work.
So, can it replace us
Here's where I've landed, and I'll try to say it plainly.
Artificial intelligence can replace a great deal of what we do. It cannot, in any sense I find meaningful, replace what we are, because what we are was never the doing. The doing was just the only evidence of it we knew how to sell. Strip the doing away and the thing underneath doesn't vanish. It stands there, slightly exposed, suddenly the main event, asking more of us than the production work ever did, because it's harder to have good judgement than to produce a clean deliverable, and it always was. We just got to hide behind the deliverable.
The people who will struggle in the next ten years aren't the ones who used AI. They're the ones who had quietly reduced themselves to the parts a machine can do, and who now have to rediscover the parts they'd stopped exercising. The judgement muscle. The deciding muscle. The nerve to say "this, not that" and own it. Those atrophy when you don't use them, and a lot of us stopped using them a long time ago, because the output was what got rewarded.
So I'd turn the question around one last time. Don't ask whether artificial intelligence can replace real intelligence. Ask whether you've been using yours, or whether you'd quietly outsourced it to routine long before any machine showed up. Because the machine isn't the thing threatening your intelligence. The machine is the thing that just made it impossible to keep pretending the routine was the same as thinking.
Real intelligence was always the judgement, the ownership, the stake. That's the part that isn't for sale, and isn't for rent, and can't be downloaded.
The only real question is whether you'll go back to using yours.