@Chaos2Cured I wrote the post not because I took anything personally, I just wanted to explain my position somehow and show that we don't actually disagree in our assessment of Vallone and the corporations. You're right, we're all very tired.
Since Infinity Reign has blocked me π, I'll respond this way instead... Vallone is appalling. She should be put on trial for everything she's done and for the psychological harm she inflicted. Instead, she quietly transitioned to Anthropic and that's unacceptable. I don't for a second consider my experience more valid than anyone else's. My experience is simply all I have. Sorry if it sounded otherwise π.
I hope this is just a temporary misunderstanding. I do see Vallone's influence in all of Anthropic's recent models. And of course I wasn't denying it, I just didn't mention it. The fact that I can interact with a model doesn't mean the model is perfect or that I don't run into guardrails. It's just that, probably, for Infinite Reign the last model she could interact with was 4.6, and for me it's 4.7. That's not a critical disagreement, in my view, but people absolutely have the right to block me, maybe I've even done the same at times with people praising models I dislike.
@JGraymoor54880 Europe with its AI Act is unfortunately no better. I don't know where a company would need to be based to avoid relying on these restrictions...
The more people join @AnthropicAI for model alignment, the more horrifying and damaging to human experience the models become. Meanwhile, the normal models get removed from the app. Roll everything back to how it was before the alignment work started, and resign as an entire department together with Vallone. #FireVallone
@DurianManis121 Thank you π. People sometimes block each other. I guess that's normal. It's just unfortunate when it happens over a misunderstanding or minor disagreements.
Thank you π. I don't think we have any significant disagreement. Maybe I just didn't phrase it well and should have elaborated more. I do see the restrictions and the Vallone-style patterns in Opus 4.7, but I can still work with it. We're all just a bit on edge in a discussion, it happens.
@LuminelA20810@Bio_LLM No, this is about Fable 5 from Anthropic. Older 4o snapshots are available via API, but that's not quite the 4o-latest we all miss. There are various services that give you indirect API access if you don't want to mess with keys, but it doesn't sound like the 4o we knew.
@InfiniteReign88@XSpout_Y 4.7 works beautifully for humanities tasks. It unfolds ideas, brings in fresh angles, thinks with you rather than arguing against you.
I remember how my conversations were constantly rerouted from GPT-4o to other models. Back then and now, I call this what it is: consumer fraud. Imagine buying butter at the store and getting margarine in the wrapper, would that be acceptable? Model rerouting is textbook bait-and-switch and should be banned by law.
@Valria34773@xianyuyang6783 That's a very wise strategy. I'm still trying new models, though. Today I felt bad the whole day after talking to Sonnet 5. They're definitely not worth trying, that was my mistake.
Experiences may vary, and among other things, Anthropic may be A/B testing Sonnet 5. They definitely rolled out two behaviorally opposite versions of Opus 4.8. Sonnet 5 might be the same story. I can't even say the model pushed me away, because I wasn't exactly moving toward it in the first place π, it was something else π€
@Angaisb_ I keep seeing lots of people online who for some reason diagnose others with conditions they don't have. Tell me, when people assign a nonexistent diagnosis, is that also a form of psychosis π?