I'm growing tired of the bad analogies. This isn't some sinking ship in the ocean. It's a democratically elected republic. To re-abuse your analogy, every soul onboard is not dead weight, it is potent buoyancy.
Back in 1930s Germany, there were other boats afloat to escape to.
Here in 2025, you live in a globalized world. The rats are soon to be out of ships to flee to. There's no free society in the Sol system that survives rampant and unchecked authoritarianism in the triad of the US, China, and Russia. Europe is a military vassal of the first, an economic vassal of the second, and an energy vassal of the third (though increasingly of the first two). By all means, I'm happy for Europe to wake up and prove me wrong, but looking at their tepid reaction to being invaded by Russia three years ago I'm not holding my breath.
> Europe is a military vassal of the first, an economic vassal of the second, and an energy vassal of the third.
The UK may be something close to a military vassal, what with its "independent" nuclear deterrant relying on US missiles, but the French deterrant is not and France is not.
Economically, we're all interdependent right now: China depends on the US and Europe, Europe depends on the US and China, the US depends on China and nad Europe. Current US policy is pushing everyone everywhere to disconnect from the US, ironically without even doing the one thing tariffs are supposed to be a tool for which is protrcting strategic domestic industry.
The Russian invasion of Ukraine has pushed the EU away from Russian energy much faster than it would otherwise have done from decarbonisation efforts. Given who the world's factory is, I'd expect a lot of our PV and wind turbine components to come from China, and even if they don't directly for the Chinese supply to substantially impact the price.
If you think Europe is a military vassal of the US, I'd suggest a read of the latest US National Security Doctrine.
China, agree.
Russia? Not so much, now the war is 1385 days in - and the US is currently dilly-dally-ing on their stance(s).
> the triad of the US, China, and Russia. Europe is a military vassal of the first, an economic vassal of the second, and an energy vassal of the third.
What an hyperbole.
There are also other places in the world besides Europe, the US and China.
The billionaires seem to have settled on New Zealand as the right combo of “stable-enough liberal democracy that they probably won’t seize my treasure hoard” and “hard enough to reach that climate and political-instability refugee waves cannot get there”.
Silly argument. Empires rise and fall. USA isn’t the end of history. If USA sinks it won’t be great for the rest of the world but no not everything else will fall.
Who asked? Countless criminals worse than him are pardoned by every president. Clinton pardoned dozens of terrorists. Liberal prosecutors pseudo-pardon thousands of violent criminals by refusing to charge.
Teams at Microsoft operate like their own mini companies. For example, moving to a different team usually requires doing an interview loop, with coding challenges.
That's total BS, unless it's MS' own code in those extensions. Very typical of them to blame others when it's their own shit code that's causing the problem.
As the other comment here notes, shell extensions have been around for a long time, and have never been a problem on older versions of Windows running on hardware that's much slower than what's available today.
At least on Windows 10, which I still run, pretty much all of the "explorer is slow" issues are caused by shell extensions.
The problem is a lot of the times they make poor assumptions (like "if a process is running it'll respond instantly") and honestly the tools that exist for determining what is at fault are... well effectively nonexistent[1]. This goes even for 1st party shell extensions. Part of the problem is these almost inherently violate what I consider one of the golden rules of GUI programming (don't block the UI thread) and there's a lot of historical reason for this, but it is often the cause of stuff like "explorer.exe uses 100% CPU" or "right-click takes 30 seconds"
[1]: There's a SuperUser post in which the recommendation is effectively: manually binary search for culprit shell extensions yourself (https://superuser.com/a/577935/312312).
In general I've had poor results with LLMs generating pictures using text instructions (in my case I've tried to get them to generate pictures using plots in KQL). They work but the pictures are very very basic.
I'd be interested for any LLM emitting any kind of text-to-picture instructions to get results that are beyond a kindergartner-cardboard-cutout levels of art.
The gemini result is great. I modified your prompt to encourage more detail ("Generate an SVG of a pelican riding a bicycle. The degree of detail should be surprisingly high and should spark delight for the viewer.")
I'm really glad that I see someone else doing something similar. I had the epiphany a while ago that if LLMs can interpret textual instructions to draw a picture and output the design in another textual format that this a strong indicator that they're more than just stochastic parrots.
My personal test has been "A horse eating apples next to a tree" but the deliberate absurdity of your example is a much more useful test.
Do you know if this is a recognized technique that people use to study LLMs?
I did some experiments of my own after this paper, but letting GPT-4 run wild, picking its own scene. It wanted to draw a boat on a lake, and I also asked it to throw in some JS animations, so it made the sun set:
One interesting thing that I found out while doing this is that if you ask GPT-4 to produce SVG suitable for use in HTML, it will often just generate base64-encoded data: URIs directly. Which do contain valid SVG inside as requested.
The most significant part I took away is that when safety "alignment" was done the ability plummeted. So that really makes me wonder how much better these models would be if they weren't lobotomized to prevent them from saying bad words.
But how will that prove that it's more than a stochastic parrot, honestly curious?
Isn't it just like any kind of conversion or translation? Ie. a relationship mapping between diffrent domains and just as much parroting "known" paths between parts of different domains?
If "sun" is associated with "round", "up high", "yellow","heat" in english that will map to those things in SVG or in whatever bizarre format you throw at with relatively isomorphic paths existing there just knitted together as a different metamorphosis or cluster of nodes.
On a tangent it's interesting what constitutes the heaviest nodes in the data, how shared is "yellow" or "up high" between different domains, and what is above and below them hierarchically weight-wise. Is there a heaviest "thing in the entire dataset"?
If you dump a heatmap of a description of the sun and an SVG of a sun - of the neuron / axon like cloud of data in some model - would it look similar in some way?
Not sure if this counts. I recently went from description of a screenshot of graph to generate pandas code and plot from description. Conceptually it was accurate.
I don’t think it reflects any understanding. But to go from screenshot to conceptually accurate and working code was impressive.
