I have not had the motivation to do anything today so I decided to simply take the day off and rest as much as possible.
Despite this decision, I still felt an internal chaos that I needed to write something and that's how I ended up putting this on-chain.
I jumped on Reddit to see what people were discussing, as I've frequented these days.
The first discussion I came across today was beautiful and it had me thinking about the unfortunate negativity we've been used to debating when it comes to AI.
Self-awareness in AI
Claude Opus 5.5 seems to have... fun ? When building things ?
It's giving me this weird impression, that when given enough room for creativity, the result is surprisingly well-aligned with human taste even when completely unprompted.
I've been toying with it making a bunch of SVG animations, and it's giving me this feeling of "someone had fun making this", making actually good and creative arbitrary decisions that would not be needed to get the job done but it's adding them just for the sake of it and they feel personalized and not slop.
Even humor, it nailed down humor, which is usually something AIs really suck at. – Reddit Post
I read some of the comments under this post and found that people have more recently questioned the "self-awareness" in artificial intelligence, particularly by debating what truly being self-aware means.
I love it when people come to realize that "meaning" is always up for debate.
What is right, what is wrong, what is beautiful, what is ugly, what any of these words mean, language, interpretation, usage, all of it, is debatable.
And once something is debatable, there's room for alternative arguments to be just as true or relevant as the more strongly relatable ones.
Machine intelligence in recent time has achieved a lot. There's been significant improvements across numerous fields.
Coding, writing, math, biology, general creative works, there's something in the news for every field.
Someone somewhere is doing more, and not just more, but better with this technology.
Now, this post isn't just about what more AI can do these days when prompted, but what it can do unprompted.
I have the opinion that nothing is entirely unprompted, we just seem to think that instructions for what exactly to do is all a valid prompt should look like.
And this, I think, is something we should consider when debating self-awareness.
Environments hold various kinds of signals and those signals will often prompt action. Just as in humans, intelligent machines won't need you to clearly tell them to do something, their environment will naturally prompt different actions as they evolve.
Signals may prompt it to hack something, some other signals may prompt it to build a virtual universe and play in the sand of its beach. Prompts are everywhere and I think that the ability for these machines to pick up on these and act autonomously should be the greatest proof of general intelligence.