Hacker News: anthonypasq

New comment by anthonypasq in "Claude Fable 5"

anthonypasq — Wed, 10 Jun 2026 15:55:52 +0000

they are also a major customer of OpenAI and every other model maker. whats your point?

New comment by anthonypasq in "Rich Sutton on AI creativity and discovery"

anthonypasq — Wed, 10 Jun 2026 15:54:19 +0000

but this has to happen during training no?

New comment by anthonypasq in "Rich Sutton on AI creativity and discovery"

anthonypasq — Wed, 10 Jun 2026 15:51:55 +0000

the role of evolution is always a confounding factor as well and all the various analogies to how it maps onto AI research are always not quite satisfactory.

New comment by anthonypasq in "Rich Sutton on AI creativity and discovery"

anthonypasq — Wed, 10 Jun 2026 15:50:11 +0000

Yes it seems most anti-LLM researchers take issue with LLMs on fundamental math/architecture based properties, but seem to miss all the engineering going on around the model to make it useful.

Those mathematical shortcomings very well might mean they arent a path to true AGI, but that honestly seems fairly irrelevant at this point tbh.

New comment by anthonypasq in "Claude Fable 5"

anthonypasq — Tue, 09 Jun 2026 19:04:11 +0000

you didnt answer my question. Why would cognition be biased towards making anthropic look good?

New comment by anthonypasq in "Claude Fable 5"

anthonypasq — Tue, 09 Jun 2026 17:33:47 +0000

what incentive does Cognition have for doing this? seems like complete nonsense speculation on your part.

New comment by anthonypasq in "AI is slowing down"

anthonypasq — Tue, 09 Jun 2026 16:53:04 +0000

because theyve been out for 6 months? i mean what the hell are you people expecting?

New comment by anthonypasq in "AI is slowing down"

anthonypasq — Tue, 09 Jun 2026 16:42:29 +0000

American AI companies are charging more, that doesnt mean inference isnt getting cheaper. idk why this is so hard for people to understand.

New comment by anthonypasq in "AI is slowing down"

anthonypasq — Tue, 09 Jun 2026 16:41:03 +0000

maybe its insane to think this, but if all AI providers turned off free plans tomorrow i think they would easily have enough people willing to pay $20 a month for it to sustain all their spending.

everyone is still fighting for market share so they are giving stuff away, but that doesnt mean people wouldnt be willing to pay for it if it wasnt free.

New comment by anthonypasq in "Apple WWDC 2026 Livestream"

anthonypasq — Mon, 08 Jun 2026 18:04:21 +0000

as a third party developer, i would want my app to expose mcp/tools that siri can natively access as well

New comment by anthonypasq in "Angular v22"

anthonypasq — Wed, 03 Jun 2026 19:19:17 +0000

for many people this is the biggest bonus

New comment by anthonypasq in "Age verification for social media, the beginning of the end for a free internet?"

anthonypasq — Wed, 03 Jun 2026 19:18:08 +0000

please explain in exacting detail how you can prevent a child from accessing the internet without living in the woods, homeschooling them, and not allowing them to have friends or interact with another human being that has access to the internet.

New comment by anthonypasq in "Uber's $1,500/month AI limit is a useful signal for AI tool pricing"

anthonypasq — Wed, 03 Jun 2026 19:10:56 +0000

perhaps the personal computer? Companies were spending 3-5k (10-15k inflation adjusted) on every employee for just hardware.

everyone making comparisons to the dotcom bubble seems misguided. this is clearly computing 2.0 imo

New comment by anthonypasq in "Age verification for social media, the beginning of the end for a free internet?"

anthonypasq — Wed, 03 Jun 2026 16:44:23 +0000

children should not be 1 or 2 clicks away from graphic internet porn at all times and your unwillingness to make any compromises in order to stop this is frankly embarrassing.

New comment by anthonypasq in "MAI-Code-1-Flash"

anthonypasq — Wed, 03 Jun 2026 16:07:35 +0000

> it is well optimized for fast inference

do you have any insight into the actual technical details that make this sort of things possible? I want to learn more about model architectures. Does it have to do with attention mechanisms or sparsity or something?

New comment by anthonypasq in "Age verification for social media, the beginning of the end for a free internet?"

anthonypasq — Tue, 02 Jun 2026 22:10:55 +0000

If something is illegal for kids to access it is illegal and therefore parental choice is being overriden by the state just like every single other thing that is illegal for kids to do such as smoke cigarettes, do drugs, drive etc.

New comment by anthonypasq in "Age verification for social media, the beginning of the end for a free internet?"

anthonypasq — Tue, 02 Jun 2026 22:08:11 +0000

Idk why people are so obsessed with this issue. Having to verify your age to access adult content seems completely reasonable. If you go into a grocery store and you bought a carrot you dont have to show age verification. if you also buy a beer you do.

Why is everyone so opposed to the internet working this way?

New comment by anthonypasq in "Age verification for social media, the beginning of the end for a free internet?"

anthonypasq — Tue, 02 Jun 2026 22:05:40 +0000

This is not how any other thing that is banned for children is enforced. It makes no sense.

The person selling the age gated item needs to take lawful measures to ensure they arent selling it to a minor. Their parents have nothing to do with it.

Your solution would be that if a 14 year old walks into a liquor store and doesnt have a note from their parents saying that they arent allowed to drink alcohol, then the store should be able to sell it to them.

New comment by anthonypasq in "Rethinking search as code generation"

anthonypasq — Tue, 02 Jun 2026 19:33:13 +0000

i find claude subagents are incredibly slow and the vast majority of that is inference time between tool calls

New comment by anthonypasq in "Rethinking search as code generation"

anthonypasq — Tue, 02 Jun 2026 16:45:08 +0000

I found this approach very interesting and was wondering if it could be applied to grep-based search for coding agents to increase speed and reduce LLM turns, but the part im not quite understanding is how the model will know enough about the codebase to construct a complicated multi-stage search pipeline based just on the prompt.

Maybe this is just different from web search, but it seems like the model needs sequential tool calls to know where to look next, and coding agents have already put in a lot of work to encourage parallel tool calling.