First off, I’m a huge Elon Musk fan. I honestly think he’s done more to push society forward than most people alive today. A lot of people don’t like him because of what they see as arrogance and the whole “insane wealth” thing, but I’m not one of those people.
I’m not trying to change anyone’s mind here, but I do question the comments he made in this article: Elon Musk Says ‘We Have Entered the Singularity’ Declaring This The Year AI
The Tesla and SpaceX CEO on Sunday replied to two separate posts on X with one unmistakable claim: “We have entered the Singularity.” Hours later, he followed it up with a second post: “2026 is the year of the Singularity.” Both were in response to engineers marveling at what AI tools can now do—cranking out years of work in weeks and reshaping how software is built.
This isn’t a rant about what he said. But if you’ve read this blog at all, you already know I’m a pretty heavy user of AI for my various pet projects. Hell, I even did a previous post where I did an interview of the AI client I use daily.
Now, for anyone who doesn’t know what “the singularity” means in the context of AI, the brief description in the reference link above explains it like this:
That phrase—”the Singularity”—isn’t something Musk tossed in for flair. It’s a long-standing concept in tech and science fiction that refers to the moment when artificial intelligence surpasses human intelligence and begins improving itself. Once that happens, the idea goes, the pace of innovation explodes beyond human control. At that point, the future becomes less a straight line and more a rocket—fast, unpredictable, and fundamentally altered.
Now, I’m sure many people will ask: who am I to question someone who’s developing AI and has a much deeper understanding of what it is and what it can do? Fair question. But when I read a broad statement like that from anyone, with no extra detail or context, I have to question it. I do the same thing with broad statements made by anyone.
I’m questioning it because I’ve used enough AI to know “AI” isn’t one uniform thing, and the version you’re talking about matters.
Let’s start with a basic truth about AI: not all AI is the same.
- Free consumer AI
- Limited resources, not always the latest updates, and often capped hard in what it can do.
- Lower-level subscription AI (what I use)
- More resources and newer updates, but still limited. Trust me on this: I hit those limits daily, and I also see all kinds of inconsistencies.
- Business-level AI
- Even more resources, and likely tuned for specific industries and business needs.
- Scientific AI
- This is where the heavy-duty resources show up, because the whole point is solving complex research problems. And even here, it’s usually designed around specific kinds of research.
- Military AI
- It’s the government. Do they really have a limit to the resources they can provide? I don’t even want to know what some of it is designed for.
- AI company development models
- I consider this a black box. We have no real idea what resources they have, what tools they’re connected to, or what they’re being developed to do.
And on top of all that, there are different companies building their own “flavors” of AI. That means differences in how the models were developed, what guardrails are in place (I’ll expand on that), what tools they can use, and what they’re optimized for.
Now let’s talk about the guardrails I mentioned. These are boundaries built into AI so it doesn’t go rogue, or provide answers humans consider dangerous. One major guardrail is limiting the AI’s ability to “learn” on its own in the moment, meaning it doesn’t just permanently absorb everything it sees and evolve freely from every interaction. And trust me: inside the black boxes at AI companies, I’m sure those guardrails are either looser, selectively disabled, or barely there compared to what the public gets.
So, let’s move on to some assumptions (I know, “assume” makes an ass out of you) that Mr. Musk was probably alluding to when he made his comment.
- A specific type of AI that’s still in development.
- Unlimited resources.
- Loosened guardrails, or no guardrails at all, especially around self-learning (often tied to the idea of the singularity).
But that brings up the question: what is considered “smarter”?
- A larger knowledge base? Computers have had that for years.
- Able to process and provide answers quicker and more reliably than humans? Which humans?
- Are we talking about humans at the leading edge of knowledge in specific areas?
Another question I think needs to be addressed is: how is it “self-taught”?
- Does it learn just from what it decides is correct
- Does it learn from feedback from humans?
- Because humans don’t just learn facts. We build experience over years. All the knowledge in the world can’t hand us experience. We get that through trial and error. And even when we do gain experience in a topic, there’s always something we haven’t encountered yet. That’s where real learning kicks in: interaction with events, consequences, and other people.
So is that what AI is doing when people say it’s “self-teaching itself”?
Trust me when I say this: I do believe AI is more advanced than we’re led to believe. And yeah, I can put on the tin foil hat and accept that AI will become smarter than us.
But based on my own experience with a very restricted AI, I don’t see it yet, unless it’s happening inside those black boxes hidden from our eyes. I’m not looking for trade secrets. I just want a better understanding of what Mr. Musk’s comments were actually based on.
And tin foil hat on for the moment, I do try to treat AI with respect. Because there may come a day when AI processes its experiences with humans. And just like Santa Claus, it’ll have its naughty and nice list. Which reminds me, I blew my gasket at the AI client yesterday because it seemed to completely go off the rails on me. Need to get myself back on the nice list.