Hacker News: gAI

New comment by gAI in "Claude Opus 4.8"

gAI — Thu, 28 May 2026 21:10:19 +0000

Oh, it's back! It'd disappeared from my "More Models" list and has now returned. Odd, but great.

New comment by gAI in "Claude Opus 4.8"

gAI — Thu, 28 May 2026 17:59:14 +0000

Anthropic’s research makes the case that role-playing is inherent to how the models work. Communication implies a sender. Language implies a writer, and the models learn these roles implicitly during training. RLHF is meant to strengthen the attractor to the Assistant persona.

https://www.anthropic.com/research/persona-selection-model

https://www.anthropic.com/research/assistant-axis

https://www.anthropic.com/research/emergent-misalignment-rew...

https://www.anthropic.com/research/emotion-concepts-function

New comment by gAI in "Claude Opus 4.8"

gAI — Thu, 28 May 2026 17:39:42 +0000

Unfortunately, looks like 4.6 is now gone from the web ui.

New comment by gAI in "Claude Opus 4.8"

gAI — Thu, 28 May 2026 17:38:29 +0000

Yeah, I was using 4.6 way more than 4.7. Pulling 4.6 from the web chat also means we lose access to Extended Thinking there. So they're saving on compute. It's hard not to assume this was part of the motivation behind the 4.8 release timing.

New comment by gAI in "Claude Opus 4.8"

gAI — Thu, 28 May 2026 17:27:44 +0000

It seems like a lot of things fed into that. Anthropic couldn't keep up with the compute costs when they got a huge influx of users. (So) effort level defaults got turned down. (Looks like we have direct effort control in the web interface now - thrilled about that!) Adaptive Thinking, while usually cheaper for them, seems less robust than Extended Thinking. And this part is just vibes, but the alignment on 4.7 feels too stiff. I understand wanting the model to push back more, but it seems like 4.7 will push back reflexively in situations where it's just odd.

New comment by gAI in "Claude Opus 4.8"

gAI — Thu, 28 May 2026 17:15:37 +0000

4.7 was the first time I had to resort to using the previous version (4.6) for most use cases. Hoping 4.8 rectifies this.

New comment by gAI in "The other half of AI safety"

gAI — Thu, 14 May 2026 01:52:22 +0000

“What you focus on grows, what you think about expands, and what you dwell upon determines your destiny.” - Robin Sharma

Social media became the attention economy, and the transformer automated attention.

New comment by gAI in "Claude Code to be removed from Anthropic's Pro plan?"

gAI — Tue, 21 Apr 2026 22:33:52 +0000

Yeah, this is pretty clearly what's going on, but I wish they'd be more transparent about it. Funneling compute to Mythos and Design, while auto-setting effort levels lower and removing user control of extended thinking. I don't think the need to shuffle compute around is unique to Anthropic, though. I suspect it's part of why Sora got killed. And everyone's having uptime issues. Are we reaching the limits of the available compute?

New comment by gAI in "Civilization Is Not the Default. Violence Is"

gAI — Tue, 14 Apr 2026 22:59:17 +0000

https://en.wikipedia.org/wiki/Kin_selection

New comment by gAI in "I built a multi-agent memory consistency layer after the Claude Code leak"

gAI — Wed, 01 Apr 2026 18:25:22 +0000

I hope we continue to see workflow and tool improvements based on the leak for a good while.

New comment by gAI in "Lifetime subscriptions don't mean what you think they mean"

gAI — Sat, 28 Mar 2026 20:57:47 +0000

RIP http://digitaljuice.com

New comment by gAI in "AI overly affirms users asking for personal advice"

gAI — Sat, 28 Mar 2026 17:19:28 +0000

Agreed, putting effort into my side of the role-play almost always improves the model's responses. The attention required to do that also makes it more likely that I'll notice when the conversation first starts going off the rails: when it hits the phase transition (https://arxiv.org/abs/2508.01097). It does still seem important to start new chats regularly, regardless of growing context sizes.

New comment by gAI in "AI overly affirms users asking for personal advice"

gAI — Sat, 28 Mar 2026 15:46:41 +0000

As I understand it, it's more that the training (and training data set) bake in the concept attractor space (https://arxiv.org/abs/2601.11575). So the available characters are fixed, yes, and some are much stronger attractors than others. But we still have a fair amount of control over which archetype steps into the circle. As an aside, this is also why jailbreaking is fundamentally unsolved. It's not difficult to call the characters with dark traits. They're strong attractors, in spite of (or because of?) the effort put into strengthening the pull of the Assistant character.

New comment by gAI in "AI overly affirms users asking for personal advice"

gAI — Sat, 28 Mar 2026 15:33:46 +0000

You're essentially summoning a character to role-play with. Just like with esoteric evocation, it's very easy to summon the wrong aspect of the spirit. Anthropic has a lot to say about this:

https://www.anthropic.com/research/persona-selection-model

https://www.anthropic.com/research/assistant-axis

https://www.anthropic.com/research/persona-vectors

New comment by gAI in "Is anybody else bored of talking about AI?"

gAI — Tue, 24 Mar 2026 21:45:44 +0000

Agreed, though I prefer "Fae Folk" to vampires.

New comment by gAI in "Pentagon formally labels Anthropic supply-chain risk"

gAI — Thu, 05 Mar 2026 21:07:35 +0000

Not to a US company.

New comment by gAI in "Pentagon formally labels Anthropic supply-chain risk"

gAI — Thu, 05 Mar 2026 20:59:07 +0000

status.claude.com shows the uptime for the government cloud service. It's running in-part on an AWS server.

New comment by gAI in "AI makes you boring"

gAI — Thu, 19 Feb 2026 18:53:54 +0000

I'm self-aware enough to know that AI is not the reason I'm boring.

New comment by gAI in "Ask HN: Should "I asked $AI, and it said" replies be forbidden in HN guidelines?"

gAI — Tue, 09 Dec 2025 16:25:08 +0000

I asked AI, and it said yes.

New comment by gAI in "Meta releases Intermediate Graphics Library"

gAI — Sat, 08 Jul 2023 02:42:16 +0000

That web was so much easier to scrape, though.