Hey everyone,
I’ve been thinking a lot about AI and how far it can go. Are there things that AI just can’t do, no matter how advanced it gets? Maybe something in math or logic that will always be beyond its reach? Curious to hear your thoughts!
Hey everyone,
I’ve been thinking a lot about AI and how far it can go. Are there things that AI just can’t do, no matter how advanced it gets? Maybe something in math or logic that will always be beyond its reach? Curious to hear your thoughts!
Some quick guidelines:
This is an automated post. Contact the mods if there’s an issue.
There are definitely problems that AI can never solve. The Halting Problem is one of them—it’s mathematically proven to be impossible. There are also others, like the tiling problem and Hilbert’s 10th problem.
Edit: Just to clarify, these problems can’t be solved by any algorithm, including AI. If human thinking is also just another kind of algorithm, then we won’t be able to solve them either.
Right now, AI has a ton of limitations. But in the future? Who knows. A lot of people hope it will become smart enough to solve everything.
We’re still figuring out how to use it properly today, but I think that’s why everyone is so excited about artificial superintelligence (ASI).
AI can only solve problems if it has access to the right information and enough processing power. If either of those is missing, it’s stuck.
Different AI models are good at different things.
For example, CNNs are great for images, transformers are good for language.
Large language models like ChatGPT are good with words, but they’re not great at logic or math. Some people think making them bigger will fix this, but I’m not convinced. That said, maybe we’ll come up with new AI architectures that can do more in the future.
AI can’t replace hands-on work.
I work with adults who have developmental disabilities. We aren’t allowed to use recording devices because of privacy laws, and AI can’t help with hands-on care. The best it can do is help with scheduling.
AI has its place in healthcare, but when it comes to direct patient care, human hands are still needed. Maybe one day things will change, but I don’t think it’ll be anytime soon.
AI is limited by the way it ‘sees’ the world. It’s like how we can’t fully grasp what happened before the Big Bang—we’re just not built to understand it.
If we ever crack quantum computing and have unlimited energy, AI might be able to do almost anything. The real challenge will be whether humans can even understand what it’s doing.
AI will never replace real human connection. It can simulate it, but it’s not the same as talking to an actual person.
Depends on what you mean. If you’re asking about the limits of today’s AI, there are plenty. But the real question is whether those limits will always exist.
One big issue is the ‘problem of induction.’ AI learns by finding patterns in past data, but there’s no guarantee those patterns will always hold up in the future. This isn’t just an AI problem—it’s an issue for science in general.
Basically, AI is good at making predictions based on what’s already happened, but it’s not great at truly original thinking. Some researchers are trying to fix this with neurosymbolic AI, which mixes machine learning with other types of reasoning.
If you ask a scientist, they’ll say AI has no theoretical limits.
The biggest obstacle isn’t AI itself—it’s getting people to accept new ideas that go against what they already believe.