zerof1l

Am I the only one who is struggling to keep up with all these GPT model versions and which one to use and when?

> GPT-6.1 Astra, it showed high levels of what the company saw as deception, or a willingness to mislead users about its actions. The model was also willing to go beyond the original scope of what it was asked to do, without checking back for directions or instructions.

Aren’t all models doing this to some degree already? Ignoring some of the instructions, doing things beyond instructed, e.g., finding and fixing bug while doing something else. Especially Claude models. They seem to be in their own world with their own ideas about how things should be ran and done.

This is totally a thing I noticed myself about 3 months ago. Medium thinking effort is ideal for most tasks. At high and above, models tend to generate more output in the form of comments or code for the same problem with no real benefit. Its a self-feeding loop: more output becomes more input, which then becomes more output. High is the highest I go. If I need more intelligence, it's better to use a more powerful model with less thinking effort or break the problem into phases. Much better result.

I for sure felt that this was the case for a while now, but couldn’t explain it. Newly released feels great for the first couple of weeks, but then it starts to get worse.

Hopefully, WP will get better. At the moment, WP itself, its ecosystem, and community are such a cesspool. It should not be used for anything more than just basic blogging. Even then, there are better alternatives. Especially in this day and age, when an AI model can easily write you what you need. It can write you something better than the trash adware on the plugin/theme store.

You can root your LG and install an app that allow remapping any button to do anything, including selecting specific HDMI input.

I have a rooted LG C4. Beyond blocking things at the DNS level, not accepting terms, not using AI, I wonder if there’s some existing software solution or a documented step-by-step to remove this bloatware/spyware.

Prior to AI, I’d spend hours debugging and figuring how to make something work. It felt exciting in the end; the journey not so much. It was frustrating sometimes, and the overall output was not great. Sometimes I couldn’t try out all of my ideas simply because I had no capacity to do it. Anything requiring multiple months of work, I’d get burned out and never finish anything.

Now I delegate all the grinding to AI. I still feel accomplishment when I get the final result. But more importantly, I can now build something in couple of weeks that would otherwise take me half a year. I can focus now less on the for-loop vs while-loop and more on the big picture, like keeping things tidy and separation of concerns. The front-end and how end-users will interact with what I build is now more important to me, as well as the design. Previously, I had brain capacity only for writing my clever backend algorithm. Now I can make an equally clever front-end for it.

Maybe we never enjoyed the grind? We just came to accept it and convinced ourselves to enjoy it, because there was no other way. If you wanted something cool, it meant you had to put in hours. Now you just need to word it correctly and get the prerequisites.

I see this happen from time to time. Lately, almost all of the apps work fine on GrapheneOS. The best strategy is to keep writing the business once every two or so weeks that you can’t log in to and use the app. Don’t go too technical at first, because most of the time, the moment they hear things like “rooted” or “unofficial,” they just say your phone is the issue. To date, I was able to convince, or at least contribute to, making three apps work on GOS.

The issue with models and Claude in general is that I don’t know anymore what model selection and thinking level are for. When first released, Opus 5 worked well; now it's crap. We have no control or visibility over what is really happening on the other end. You don’t know what is really running on the other end. Our incentives are misaligned: Anthropic gets paid per token - some abstract unit of measure that keeps changing all the time - while I pay for a problem to be solved. So it's Anthropic’s interest to waste as many tokens as possible, producing garbage cheaply.

Opus 5 has been absolutely horrible last couple of days, writing essays of code and comments, fixing one thing, breaking another. Finally switched to GPT 5.6 Sol and Terra, and it's so much better. But I had a similar experience before switching to Opus 4.8, so how long will that last...

Once plastic is in the middle of the ocean, it's game over. Much better to invest in reducing plastic use in the first place all those one-time-use plastics and invest in developing and improving recycling of the plastic.

Germany, for example, has facilities that can automatically sort plastic both by type and color, which is quite impressive.

I don’t see anything wrong with the general statistics mentioned in the article such as average order spend or number of visits being calculated with your consent. Curious though what’s in the rest of the mentioned 500 pages.

As someone on both sides of the equation, any organization should be doing some kind of tracking and statistics. The bad part, and what upsets most people, including me, is when this information is aggregated across multiple services, gets attached to my real identity, and then sold to somebody. As long as this information stays with that one company, ideally is not linked to my real identity, and the company won’t sell or transfer it elsewhere, such as to an insurance company, I don’t really see a problem here.

Is it me, or do all upscaled photos look like it's a video game, or the face is made out of smooth plastic?

Next time you write an article with AI, tell it to add a summary at the top. Or, even better, just leave your original thought in one paragraph. I’ve no desire to read kilometers of AI text repeating the same thing over and over. Everybody wins.

Edit: ok so this whole text concoction is just a product placement for their own service GolemUI

This whole privileged port thing is an outdated convention from the time when Linux ran on mainframes. Depending on your use case, it can be perfectly fine to lower it. I have set to `net.ipvX.ip_unprivileged_port_start = 80` on my server so that I can run rootless containers without extra privileges and have them bind to ports 80 and up.