jari_mustonen

> Qwen barely moved toward Opus 4.8 in the earlier experiment, but moved by +20.58 points toward GPT-5.5 Pro here, including a large effect on the private synthetic puzzles. The data suggest that Qwen may have learned from GPT-5.5 Pro, or from a closely related GPT model, rather than from Opus.

How does this suggest anyting of the sorts?

In fact, they are also selling drones to Ukraine.

"My primary concern is the risk that authoritarian governments—not solely the Chinese Communist Party (CCP), although the CCP is clearly the most capable threat—build AI models that are more powerful than those built by the US, and use them to achieve permanent military superiority"

I see this sentiment a lot. China is not perfect by any stretch of the imagination but since 1979 China has not participated in a single war or supported hostile regimen change operations.

I think Amodei's mistake is to take it granted that USA is a good actor. Anyone can draw their own conclusions but just for reference here are some highlights starting from 1979.

Armed operations in Lebanon (1982-84), Grenada (1983), Libya (1986), the Persian Gulf (1987-88), Panama (1989-90), Iraq and Kuwait (1990-91, with no-fly zones until 2003), Somalia (1992-94), Haiti (1994), Bosnia (1995), Sudan and Afghanistan (1998), Iraq (1998), Serbia (1999), Afghanistan (2001-2021), Iraq (2003-2011, and again from 2014), Pakistan (2004-2018), Somalia (2007 to the present), Yemen (2002 to the present), Libya (2011), Syria (2014 to the present), Iran (2020 and 2025 to the present), and Venezuela and the Caribbean (2025-26).

Regime change operations in Afghanistan (1979-89), Nicaragua (1981-90), Cambodia (1980s), Angola (1985-91), Iraq (1995-98), Serbia (1999-2000), Syria (2013-17), and Venezuela (2019-2025).

The lists do not include the numerous operations by Israel which effectively is part of the same US military hegemony that Amodei is here defending.

I had a steange experience talking to Claude about a random 15-years old boys winning women’s national soccer team.

It insisted, to a level that I detemined it to be part of the post training, that it does not matter. That women’s team is better at dribbling, finishing and reading the field, it claimed. When I pushed it how it knows this, it didn’t let go. Instead it started claiming that it has personally observed this by watching the games.

Every society has its taboos and ours is no different and feminism being just one.

In my books, the chinese censorship is better. You know where they are holding their finger on the scale and not hiding it behind vague terms like ”safety”.

It seems that the US is consumed by the security state. Each and every aspect of the economy is subjugated to the need to maintain empire. Just to give some random examples: the weaponization of the banking system (kicking Russia out of swift), the semiconductor export ban to China, the TikTok ban, or the blatant over usege of tariffs.

Now they are betting that Mythos will provide them with some edge. Personally, I don't believe that Mythos is such a game changer They're just buying their own hype.

A late stage empire flailing around sacrificing everything to maintain its status.

Maybe I'm just old school but I prefer to buy the rocket business without a social network attached.

In the screen shot, "Claude Code" with "gpt-5.5" model.

Here's a note about running entirely on Huawei chips:

https://finance.yahoo.com/sectors/technology/articles/deepse...

Open Source as it gets in this space, top notch developer documentation, and prices insanely low, while delivering frontier model capabilities. So basically, this is from hackers to hackers. Loving it!

Also, note that there's zero CUDA dependency. It runs entirely on Huawei chips. In other words, Chinese ecosystem has delivered a complete AI stack. Like it or not, that's a big news. But what's there not to like when monopolies break down?

> An F-15 being shot down in Iran after weeks of strategic bombing of their anti-air defense systems is not a good sign.

Wrong. It's a great sign. We have had enough of the barrage of US aggression around the world.

Some of these are neccesary parts of LLM's. They use the content they create to direct what they are going to say. This applies to patterns like "In conclusion, ..." and what the author calls "Fractal summary". Turn them off, and the general quality of the AI thought gets lower.

There is nothing ideal about that outcome. The "regime change" people talk about is intended to look like what happened in Libya: A failed state that falls in anarchy.

The classic "Frequently Questioned Answers" is the ultimate takedown of this abomination of a language.

https://yosefk.com/c++fqa/

There are sexual predators that do pray on men and the sadistic types probably find their way to prison guards.

Stuff like this is good for infinite loops: https://gonephishing.me/shell-jacker/shell-jacker/worm_launc...

So they integrate Gemini to summarize open web pages and consolidate all your open tabs into summaries. (Open lot's of pages, then summarize them all.) You can search your history with natural language and type Gemini queries directly into the address bar.

This will give them a cognitive profile of you: reading comprehension, decision-making patterns, knowledge gaps, etc.

Scary.

So they integrate Gemini to summarize open web pages and consolidate all your open tabs into summaries. You can search your history with natural language and type Gemini queries directly into the address bar.

this will give them a cognitive profile of you: reading comprehension, decision-making patterns, knowledge gaps, etc.

Scary.

Here is the summary of key improvements made:

1. Structure & Flow

    - Decision Trees: Clear branching logic with ├── and └── notation

    - Sequential Steps: Numbered, ordered procedures instead of scattered explanations

    - Prerequisites: Explicit dependency checks before proceeding

2. AI Agent Optimizations

    - Tool Call Clarity: Exact function names and parameters

    - Binary Decisions: Clear yes/no conditions instead of ambiguous language

    - Error Handling: Specific failure conditions and next steps

    - Verification Steps: "Recheck" instructions after each fix

3. Cognitive Load Reduction

    - Reference Tables: Quick lookup for tools and purposes

    - Pattern Recognition: Common issue combinations and their solutions

    - Critical Reminders: Common AI mistakes section to prevent errors

4. Actionable Language

    - Removed verbose explanations mixed with instructions

    - Consolidated multiple documents' logic into single workflows 

    - Used imperative commands: "Check X", "If Y then Z"

    - Added immediate verification steps

Is it capable of generating white males?

Given that this has been a serious problem with Google models, I would guess it would have been a good thing to at least add one such example in the marketing material. If the marketing material is to be believed, the model really prefers black females.

It's the political bias in the training material. No surprise there.

This is really useful for Claude Code. The second opinion is often really insightful. Also, in a lot of cases, when CC can't fix a bug, Gemini can.

I think the problem is that removing factors like name, gender, or marital status does not truly make the process unbiased. These factors are only sources of bias if there is no correlation between, for example, marital status and the ability to work or some secondary characteristic that is preferable to employer such as loyalty. It can be easily hypothesized that marital status might stabilize a person or make them more likely to stay with one employer, or other traits that are preferable.

Similar examples can also be made for name and gender.

The gender bias is not primarily about LLMs but rather a reflection of the training material, which mirrors our culture. This is evident as the bias remains fairly consistent across different models.

The bias toward the first presented candidate is interesting. The effect size for this bias is larger, and while it is generally consistent across models, there is an exception: Gemini 2.0.

If things in the beginning of the prompt are considered "better", does this affect chat like interface where LLM would "weight" first messages to be more important? For example, I have some experience with Aider, where LLM seems to prefer the first version of a file that it has seen.

Of course, AI will be used, first and foremost, to push advertising messages.

Is it just me, or do others also wish they could use public transportation, go for a walk in a park, or watch an educational video without being constantly bombarded with ads at every turn?

If there’s a jurisdiction that has an environmental protection tax on display advertising, whether in the real world or online, let me know. I’ll be the first to move there.

The article mentions Jung's "collective unconscious," which is often misunderstood.

What he meant by it is that some unconscious features are collective, meaning they are genetically programmed in all people. Jung believed this also includes certain thought patterns, which can be inferred from stories. For example, he would have argued that a paragon of wisdom is typically an older man with a white beard (Gandalf and Dumbledore come to mind) because we have a genetically programmed inclination to see older men with white beards as paragons of wisdom.

Jung liked to use these kinds of methods to analyze the human psyche and its structures. Interesting guy. If anyone is interested, I recommend his collection of essays, Modern Man in Search of a Soul, as a first read.