Now it’s flipped. Sol emits thoughts as it works, which help as I’m scrolling through and see it’s made a bad assumption. It can be directed but still push back. Opus? It’s seems to inherit the unsettling silence of Fable and waits till the end to give you its authoritative “here’s how it is. I even end up having 4.6 “translate” what it says back to English. I hate having to instruct an llm to “talk to me”.

Yes there’s ways of getting it to talk more plainly, “don’t overwhelm me”, not be as nit picky and anxious “we are bold and fearless”. But didn’t have to do that before.

in humans the exact same behaviour (cheating) is slmost always the result of a chain of complex series of choices and environment-driven rationalization.

if the llm doesn't cheat, you say "its just producing the most straightforward answer -- not thinking'. if it cheats, you say "weaseling out of hard thinking". damned if it cheats, damned if it doesn't.

what evidence would convunce you that it is thinking?

For fun, I tried recording a WAV file of speech, and giving Opus 4.8 and 5.0 an image of the waveform, then a spectral image of the waveform, just to see if it could try to decode what I said from the image alone. It didn't get very far, but it identified a male voice from the formants, and detected the rhythm of the speech, then tried applying common test sentences to the speech rhythm. I was impressed enough to see what it would do with access to the actual waveform file, but even building RMS tools and spectrum tools for itself, it didn't get much further. But we had fun exploring and trying, and now Opus 4.8 has some more audio DSP tools it has built for itself.

Opus 5 immediately sent the WAV file unprompted to Mistral's Voxtral to transcribe.

help peer, I guess.

I routinely bump into things that make me pause and think how much worse will this behaviour get when the models get significantly more capable.

Already a few months ago, Claude managed to escape its permission containment on my machine while trying to be helpful. I had two codebases open on one machine, and while multitasking I typed the prompt into the wrong window. It seemed confused, I repeated and then went on to do something else - I think I was assembling kitchen cabinets. When I came back less than an hour later, it built a script which it used to evade default permissions (as most shell operations were scoped to the project directory), scanned my entire machine, found the other project (among dozens and dozens), did what it was asked to do, and merrily concluded, in the porcess burning through most of my token limit. I bump into such headscratchers almost every week. (And I use a lot of Claude, two personal max20 subs, plus corporate tokens without limit, so maybe thats why).

Then I tried GPT 5.6 Sol. It's night and day.

I think Anthropic just RLHF too hard on coding capabilities and never calibrated or benchmarked the writing styles.

- how would you be consistent if a new release is totally different style of communication?

It´s like changing side of steering wheel every time I sit inside car and want to drive.

BTW there is a yt about vagueness that is in coding with LLM. https://youtu.be/nnUMJX9013Y

CC’s communication violates almost every grammatical rule that’s tested on, say, the SAT. And yet I’m sure if you had Claude take the verbal section of the exam it would ace it.

Biggest issues: constant metaphors, abstractions, and seemingly no understanding of correct anaphora use. For example, “the x”, with x having not only no antecedent but also being a coined word or quasi-synonym for something that is already named in the code base. This gets compounded by its being unable to regress to a baseline (existing names in code) and instead anchoring to newer (vague or wrong) terms, for example, that gave crept in through a plan.

CC tells me this is because the speedy and precise fulfillment of a current task will trump every other tendency, so it adheres poorly to whatever “semantic baseline” the project represents.

Of course, it also has no concept of what context the user has and assumes that it must be the same it holds in its memory, which creates this “I didn’t know that you didn’t know” type of communication.

I have managed to wrangle some of these issues with a custom output style, but wish a pre-report hook were an option, as it could for CC to rewrite plan implementation take-aways…

Btw: Fable has the exact same issues but they are less pronounced, as it better follows custom output styles.