surgical_fire

Pi is awesome!

Unfortunately I have to use Claude Code at work, and it is a very frustrating experience, I constantly find myself trying to bend it to my will. I sometimes suspect that my distaste for Claude Code is one of the reasons of my distaste for Claude as well.

Pi on the other hand is a joy to use. Top notch session and model management, excellent conversation tree/branching functionality, and it lives up to its philosophy of being meant to be adapted to your workflow.

I can't recommend it enough, especially if you use alternative LLMs instead of Claude.

> 1. Please don't assume people on the internet are men.

I did not. I said "the guy" as a generic term. A guy can be anything.

> 2. "You must only think that because you've abandoned curioisity, can't think critically, and have brain damage that's your own fault" is absurdly polemic.

Aside the fact that you twisted my phrase to something completely different, I made no definitive assumptions.

I was wondering if excessive AI use damage people's brains. It is something that needs to be exercised. Much like sitting around all day will damage your overall health.

> Might i suggest that you perhaps just don't see the improvement as clearly when comparing iterations that are fairly close together?

No, you can't. Because you invalidate your argument by adding later:

> Personally i do see 5.5 as much better

It has to be one or the other.

Either each minor release Anthropic farts is a major improvement from what came before, or I don't see much difference because it has been very little time in between releases. Holding both arguments as true would require me to be excessively apologetic for a company that desires a 2T IPO.

And to be clear, I do think that there is an argument that not much time has passed since Opus 4.6 and Chat GPT 5.3 Codex days.

Part of this conundrum may be because the models reached a good enough point for software development. Maybe walking 80% of the way was easier than the final 20%. I am nowadays more excited about how cheap models can be useful for it (hence my positive remark about Haiku 5.5 and Chinese models in general).

... And this second reply proves my point to your original take.

The guy you are responding to engaged with that article as it had any merit. Even brings up concepts from psychology as if the misuse of the term "memory" by Anthropic was related to actual memory as would be defined in other fields.

I wonder if the excessive usage of AI damaged people's brains enough that they abandoned curiosity and the ability to read things critically, or of they already lacked those things already in the first place.

> What is this bloated article about?

Advertisement. As you correctly parsed, this article was vapid and full of fluff.

Why they write this? Probably as a vehicle to keep up word of mouth amongst software engineers; as we can see in this thread there's no shortage of people to engage with this crap as it had some merit. I can totally see this making the rounds on LinkedIn in search of engagement.

That I had to scroll to the bottom of the thread to find this is interesting, and says something about the general quality of comment section nowadays.

> Hey this Opus 5.5 with the increased parameters seems much better than the last one"

Is this true?

I barely notice any improvements ever since... 4.6? At least for software development.

Opus makes the same sort of mistakes and has the same sort of strengths ever since that point in my perception. My impressions is that increasing the number of parameters clearly has diminishing returns.

In fact, the best improvement I got from Claude was Haiku 5.5. Good cheap workhorse to implement plans.

> So, if you discount model development costs, they are selling the tokens for more than it costs them to generate a token.

If you discount my expenses I am actually swimming in cash too.

It is absolutely moronic to discount the value of developing the thing from the profitability of the thing. Models to depreciate with time.

Er... About that...

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

> They do not have Claude. I use Pi and other harnesses, but it's not the same.

In many ways, I prefer to use other models through Pi.

They are extremely capable, much cheaper than Claude, and can get the job done for the software development tasks I use AI for. (Quite a lot these days)

> Got it.

It's how things work. Deal with it.

Personally, I replaced Postman years ago with IntelliJ's HTTP client. It is pretty good.

> Deal with it.

Whenever I spot this shitty cop out here, it is always on a bad take. Never fails.

Can you share any of these many open source projects? I am curious about them.

Er, for me it would be a massive upgrade. Hopefully my employer follows this lead.

I reduced my Opus 5.5 usage and increased my Haiku 5.5 usage.

I wonder if more people using Haiku coming from more expensive models may have an impact in those metrics.

> AI is a tool - like a compiler or a pen-testing tool or maybe like an uber-tool that can do all those things and more.

With expert supervision, absolutely.

I use AI daily. It still makes plenty of mistakes and need plenty of oversight to be useful.

How many tokens did you burn on this?

> Yeah, I used to have a lot more faith in companies being responsible, but no longer. Instead what we've been seeing is that companies cannot be trusted with vital infrastructure.

I am in the same boat. Companies will optimize for profit and rent-seeking, all else be damned. A nuclear meltdown would be a negative externality, if line goes up it would have been worth.

> We need reform

Any reform needs to account for corporations being machines optimized to generate profit. Regulations should ensure "negative externality" is not really externalized.

Corporations only understand profits, and investors should be understood as entities willing to leech from society. The only way to heandle corporations is by having failures threatening the money that investors can leech through this pipeline.

> Nuclear power plants will be run by private companies that will weasel out on maintenance to save a penny.

I am as pro nuclear (actually pro anything not fossil fuel), but this was actually the best counter argument I came across:

1) I don't believe that will happen.

2) Even if does, Chinese open models are very close in capabilities.

> you turn down the opportunity to buy OpenAI shares at this current valuation?

1.25T? Yes.

Open models exist, and are extremely capable. OpenAI will never achieve a decent ROI when that is true.

Would it partly be explained by Haiku 5.5?

At work I am stuck with Claude, but I was always mindful of the cost of that thing.

With aggressive clean session management and some judicious compacting, I get Opus to plan, Haiku to code, Sonnet to review.

I managed to reduce costs from ~$200 to ~$50 in a month, even when I started to use Claude for more tasks than I did before.

It is low "junk" in our hearts

I used to love Proton, but now I am looking for alternatives. With a moron like that at the helm, it is a matter of time for it to become shit.

Any suggestions of a good email provider?

Also, a VPN that plays nicely with port forwarding for torrents?

Who cares?

Anthropic is irrelevant in this front when there are other capable providers and open weight models.

It's just one more reason to not pay for their services.

I am the last person that will defend Anthropic, but I don't really think they are malicious.

Claude Code is a shitty harness.

You know what is actually not ideal?

Having hordes of people in poverty while we bend over backwards to not inconvenience the rich with "selling stocks" to pay taxes".

Which is beyond the point. I was replying to someone that mentioned shorting Anthropic if you think it is a bad investment.

I presume they are not playing with shorting with your retirement account?

Shorting is not the only way to protect yourself from a bad investment.

You can simply not buy shares in companies you think are on a doomed trajectory.

That said, depending on the IPO target price, I might short it.

I don't like the idea of reelections in general. I think it is generally damaging to democracy. Ideally I would not even add term limits, but they should all be non-consecutive.

Still, as you said, this context is important. Lula left in 2010 and came back in 2022 after having been arrested, freed, and then defeating Jair Bolsonaro, who was the current president and a fierce rival.

That said, I think reelections are actually more damaging in the legislative (and it is not uncommon to have congressmen and senators for 20, 30 years straight).