13 cantos revisited
I've been discussing 13 cantos with ChatGPT o3 for the past couple of weeks or so — it's taking that long as I've periodically run into the prompt quota for the model, even on Plus.
tl;dr — It began well enough, but devolved rather quickly into an odd experience in which I felt like I was being dissected as if under the baleful stare of a clearly alien intelligence. ::chuckle:: (In sheer contrast to the warm, eerily human-like assays of both Claude [3.7 and earlier] and Number 1 [GPT 3.5 and 4o].)
While I did appreciate o3’s cogently analytical and acceptably literary analysis, it came across as rigid and overly structured. Users of o3 will surely recognize this in terms of the relentlessly tabular presentation of its ideas. There’s something to be said about that form — for one thing it allows for a quicker visual parsing of the content, and for another it renders o3’s ideas and analyses in a beautifully logical and rational way.
On the other hand, I miss the discursive style of the previous (non-reasoning) models, which rendered a more fluid narrative. I guess it’s the difference between reading a research paper and listening to a presentation of it.
Now: which modality results in a keener engagement with me, the interlocutor? I guess it depends on my mood at the moment — I think I’d prefer the more discursive style during Vigil, for example. [ That being the interregnum between First and Second Sleep, in the biphasic way of things. ::chuckle::: ]
As to the question of which model is the more perceptive, or more intelligent: I can honestly say the jury is out for me on this for now… I’m not sure if o3’s putatively sharper cognition is worth its reportedly heightened proclivity for hallucination (which I’ve definitely seen, and which troubles me).
We’ve only gotten through 3 canti thus far, given the aforementioned quota limitation, so I guess any conclusion about o3’s perceptivity and cognitive acumen will have to wait. Or be provisional, like this: even at only a quarter of the pieces, it’s already seen a couple of things I hadn’t considered before. Which is saying something, inasmuch as I’ve lived with this work for almost a decade now.
Not a novel observation: we’re living in truly extraordinary times, with respect to the evolution of intelligence itself.
[ thought in progress, throughout the day, interstitiae-filtered... ]
