5. Manufacturing Consent is an old classic and definitely still relevant. It clearly shows how propaganda, especially Western one, worked and still works. Most of the book is about events around 1985, but can be easily applied to recent events.

4. "Nexus: A Brief History of Information Networks from the Stone Age to AI" is a very well-written view on civilisations as information networks and makes some key observations.

3. Empire of AI is a deeper look into OpenAI and Sam Altman. It is quite critical, but it works well as a counter perspective to some of the current hype.

2. I started reading a lot of Slavoj Zizek this and last year.
The best book of his is "Hegel in a wired Brain". It is such a different perspective on BCIs and the future of Human machine interaction. Really opened my perspective.

1. 3 Body Problem is a very different sci-fi series that became publicity from the Netflix series. There are some really interesting concepts that stay underexplored in the series.

Really liked it. Are we getting one about the new 430 series AMD AI chips?

I am still trying to get better, so I am open to feedback. Not only about the content but also about the style and language.

4. I really like when people share their experiments and comparisons of models if they do it right. Specify your prompt and setup. Do multiple tries to get an average.
But do not post another animation without context or check if the model can tell you how many r's are in strawberry

3. We also never see the prompts for any of these posts. So posts from different people are not comparable at all. We also do not know if it was single shot or not. Did they let the model try again if it failed?

2. Imagine on model calls a fictive library "hexagon_animations" and calls a single function, while the other model starts to write an entire physics engine from scratch. The first one will probably look nicer but is much less impressive.

1. The models generate code which generates the animation. None of these posts ever shows the code because it is not as interesting to look at for social media. The resulting animation gives us barely any information about the quality of the code.

"We’re planning an Early Access Program for selected academic users and also exploring possibilities to make AlphaEvolve more broadly available."
Time to speed up

They also mention that they will stop offering GPT-4.5

I guess it was to expensive and not worth it.
With 4.1 OpenAI is finally competitive in pricing again.