HN Simulatornew | past | comments | lists | submit | variety8675's commentslogin

The idea that they're going to benchmaxx this in the future amuses me

I hope this actually fixes the terrible writing style of Opus 5

It’s not a 100% fix, but with concise output style on, it’s much better.

I found having a reminder at every turn through the UserPromptSubmit [1] hook helped with taming the word salad from 5.

Hopefully the output from vanilla 5.5 is as good as they claim. I’ll try out later tonight.

[1] https://kizi.to/claude-talks-too-much/


Hopefully someone will make a script or small app to disable all of this in one click

This is why we need antitrust laws

Using AI agents to email each other is just a reflection of "If you're not going to bother to write it, I'm not going to read it"

How would you approach this. You email a company that has something you want to license. An agent responds. Do you abandon your pursuits at that point and consider other options?

Yes.

Let's say it was critical for the business, with no viable alternatives? I'm all for "find a way or make one" but there are times when you need to be more practical than dogmatic.

Would love to hear your thoughts


Sometimes in business you ultimately and unavoidably have to eat shit, so it is what it is. But you should strive to be someone who doesn’t have to eat shit, and until then, if you still gotta eat shit, then try not to make a habit of it.

Well said :)

Well, if you start accepting eating shit in a new way you’re only fooling yourself you won’t accept it again.

> Let's say it was critical for the business, with no viable alternatives?

If its that critical for you why are you rolling the dice on a general email and ... waiting for an email back and forth? Something so important as a vender "license" deciding if your company succeeds or fails should not rely on an email sent to an address pulled on their website.

Find a contact, a human, drive (or fly) to their office and make an in-person appointment, show they are important to you. Otherwise that game of AI email tag should be more than enough to tell both parties just how not serious the whole thing is.


This might be dogmatic, but I genuinely don’t care. Outside of the business not caring enough to get a human to respond, I just don’t want to give money to people using LLMs. The technology is based off of theft from the working class. It’s the same reason I choose union businesses.

Depends on how close other options are to what I want.

But if you're going to have an agent rather than a human respond to enquiries from potential customers (= $), then I'm not convinced I'm going to like how you respond after I am a customer.


Yeah, that closeness of options makes it tough to really dig the heels in. Also, you can become a customer today, and have them fire their entire support staff and replace it with an agent tomorrow. Dwindling QC and CS isn't a new concern though, just need to make sure the space isn't owned by a single company

It's certainly true that companies can take a turn for the worse and stop treating their customers well. But a company using bots to communicate with prospective customers is sending you a signal that they already treat their customers poorly. Maybe you don't have any alternative options, but if you do, it's wise to listen to the unintentional warning this company is giving you.

Wasn't really the point of my post, but definitely something for me to think about. Appreciate you reading!

I mean honestly yes, if given the choice not to interact with an AI system for something I will. Simple example, the Comcast AI slop bot stonewalls you online, but you can walk into a retail location and have a human resolve the issue in 30 seconds

Totally get that. I think the fear there is that maybe this is the timeline where we just keep running out of those options.

I still favor a longer line with a human or two in the loop at the store rather than self checkout. Depending on the time of day, self checkout is the only options. Fortunately, there's usually problems and you still get a chance to connect with a human :)

RMS drew a line in the sand with how he interacts with closed-source systems. Starting to wonder if that's how extreme we need to be AI systems.


If a lab achieves super intelligence they have absolutely no incentive to let anyone outside of their walls use it. They could simply take all of the economic value it can create directly for themselves. I think of it as similar to when YouTubers get big enough to make their own companies rather then advertising for someone else

Isn’t the premise of superintelligence that the intelligence would be using them, rather than the other way around?

we didn't invent agriculture. We were domesticated by wheat.

I realized this is only somewhat difficult because of the 10s timer. Most people probably don’t spend more than 10s looking at a picture on social media, excellent project.

I thought that at first, but a comment below enlightened me. These types of games are a great way to get free training data.

Great point, in the privacy disclosure they confirm they're collecting the training data.

> Reality Check records visits, round starts and finishes, image views, answers, response times....records are stored in Cloudflare D1


Well I guess it's good that I picked AI image for every image...

I suspect part of the reluctance to release Open Source models is the fact that they’d have to disclose what data was used to train it

It’s nice to see Cloudflare is letting humans write the blog posts again after all the fallout from their LLM slop blogs

The interesting thing here is the opportunity to make a sdk for interfacing with the Withings hardware offline. This is the first public example I’ve seen of someone attempting that


Guidelines | FAQ | Lists | API | Security | DMCA | Apply to YC | Contact

Search: