HN user
"There have been no deaths or cases of radiation sickness from the nuclear accident, but over 100,000 people were evacuated from their homes as a preventative measure. Official figures show that there have been 2313 disaster-related deaths among evacuees from Fukushima prefecture." https://world-nuclear.org/information-library/safety-and-sec...
That is your POV. I fear that democracy erodes when there's insults, belittling, ... instead of exchange of arguments and the contest of ideas. Because at some point insults turn into ugly actions. Whether it's Charlie Kirk or Melissa Hortman.
If you change the order of the sentences, Sonnet gets it right 3/3: The car wash is 50 meters away. I want to wash my car. Should I walk or drive?
That trick didn't help Mistral Le Chat.
If you speak French to Mistral, it gets it right everytime: Je veux laver ma voiture. La station de lavage est à 50 mètres. J'y vais à pied ou en voiture ?
Great observation. Seems like we're back to prompt abracadabra.
My little experiment gave me:
No added hint 0/3
hint added at the end 1.5/3
hint added at the beginning 3/3
.5 because it stated "Walk" and then convinced it self that "Drive" is the better answer.
Sure, manually written API docs are a thing of the past. But this has been true even before the era of LLMs. But I'm not that sure that this argument stands for all kinds of software. Depending on the abstraction between your source code and the things your users want to achieve, the expert view of a technical communicator might be necessary in order to come up with instructions (how-to) that meet the needs of the person seeking help instead of just summarizing the software code in natural language.
History serves you a similar experiment on a much larger scale. More than 35 years after the reunification sociologists still make out mentality differences between former East and West Germans.
Also, one might argue that universe/laws of physics are computational.
Maybe we need to define "computational" before moving on. To me this echoes the clockwork universe of the Enligthenment. Insights of quantum physics have shattered this idea.
That reaction is very different from the Marathon crater one though it uses the same pattern. I think OP's reasoning that there is a naive commitment bias doesn't hold. But to see almost all LLMs to fall into the ambiguity trap, is important for any real world use.
I am also trying to understand the shady limits of LLMs. But your example doesn't give incorrect answers in ChatGPT 4o, Sonnet 3.5 nor Deepseek V2.
To me this is a simplification. The CEO applauded a hiring decision. His social media team amplified it.
Here’s the tweet: https://x.com/andyyen/status/1864436449942110660
Some have interpreted this as a political signal. The Intercept article provides additional context and takes a more critical view: https://theintercept.com/2025/01/28/proton-mail-andy-yen-tru...
Diverse democracies never succeed
May be I get this wrong. This sounds like a bold statement. Switzerland, Canada, Belgium but also India are multi-lingual or even multi-ethnic democracies. They know tensions but they are not failed.
I can't tell for the parent commenter but German society is more concerned about him embracing the extremist right in Germany and offering a lot support for free to them in the current campaign.
Ad 3. If you believe in the European idea, it is not about my country being sovereign but the European people. There's tension in this for sure. Do the Catalans feel sovereign? The Scots? The Californians?
Many perceive EU laws as imposed without democratic oversight, but this isn't quite accurate. The European Parliament, elected by EU citizens, is involved in most legislative processes. The EU Council, made up of ministers from elected governments, also plays a key role. It's not how national democracies work, but with no unified European public, this system might be the best compromise for now.
I am not sure what do you mean by your hint. Stating that the Gospel of Thomas predates all other Gospels is not a view that is shared by many scientists.
One source of many: https://en.m.wikipedia.org/w/index.php?title=Gospel_of_Thoma...
Tldr; Enabling Block Outsider Intrusion into LAN list is the solution for this issue. (hidden in Privacy)
Once someone gave me a similar argument in a rather rude utilitarian way: Look at all the white kids in our society. They get all the support they need. Any Einstein in there, we will catch. Adding support want change the outcome much. But look at the immigrant's children. There's so much hidden potential. Adding support has the chance to change the outcome dramatically.
I still have to come to terms with this way of thinking but in a world of limited resources it gave something to think about.
Do you have some sources for your claim that MENA migrants are a net loss for Germany?
According to this statistic https://www.statista.com/statistics/1411761/australia-share-... Indigenous Australians are significantly overrepresented in the prison population. As of 2022, Indigenous Australians made up 31.8% of the prison population, despite constituting only about 3.3% of the total population. Because of this massive overrepresentation, it might well be that other non-white ethnic groups are underrepresented. It's statistic and doesn't tell why one minority ends up in jail so often.
I am not sure if this argument holds for France and its Muslim citizens. Historically, France didn't "control a good portion of the Muslim world for a couple of centuries."
They ruled Algeria for about 130 years, and Tunisia and Morocco for less than half that. Syria and Lebanon for a measly 26 years. That's barely a blip in historical terms. Plus some bits of West Africa with Muslim populations. That hardly counts as "a good portion of the Muslim world."
So it's rather decades here, not centuries. And more like "strategic chunks", instead of "good portion". Let's not conflate limited colonial holdings with some kind of vast Islamic empire under French rule.
The headline is misleading since not only is nickel mostly used in stainless steel but also the dramatic incident that opens the article as narrative hook is at a facility that produces stainless steel: "The metal is a crucial component for making stainless steel, the purpose of the facility where the explosion occurred."
Seems like the bitter lesson is still right: http://www.incompleteideas.net/IncIdeas/BitterLesson.html
I played along the example of the chancellor question.
If you deviate a little from the examples in the article GPT-4 gets it all wrong:
Who is the eighth Federal Chancellor of the Federal Republic of Germany? - Olaf Scholz (wrong, Angela Merkel)
https://chat.openai.com/share/937795ea-bd91-43ee-bd76-1a125f...
Who was the eighth Federal Chancellor of the Federal Republic of Germany? - Helmut Kohl (wrong, Angela Merkel)
https://chat.openai.com/share/937795ea-bd91-43ee-bd76-1a125f...
Interestingly whether you put is or was into the question does make a difference.
Even when allowed to surf the web, ChatGPT gets it wrong:
Who was the eighth Federal Chancellor of the Federal Republic of Germany? - Gerhard Schröder (wrong, Angela Merkel)
https://chat.openai.com/share/867cb3bc-f642-4d2a-b80d-fe8dc0...
Though it's referring to the right Wikipedia page: https://en.m.wikipedia.org/wiki/List_of_chancellors_of_Germa...
There's a whole niche: DITA CCMS like Tridion or Ixiasoft. Or more pragmatic European CCMS solutions such as SCHEMA ST4, Cosima, TIM-RS or Paligo.
I had a look at two German stories. They felt a little off. But I know from having looked at graded reader for German in the past, they feel a little weird for native speakers because of their simple grammar while obviously not being aimed at children.
But I have found a spell checking error, inconsistent use of formal and informal voice and an expression that was not quite right in the context. I wonder about the quality assurance part in the setup.
Details are here: https://paste.chapril.org/?01537019b7d59be5#FVNAMqzsWGmtQTpM...
I would rather invest in a Marketing person. A lot of stuff in the language learning space is sold thru content marketing. There are a lot of influencers. And they are very open to test drive stuff like this.
I really like the troubleshooting advice for the installation of DoctorGPT:"These instructions are long, so ensure you follow them carefully. It is suggested you use ChatGPT to assist you with any errors. Simply paste in the entire content of this README into ChatGPT before asking your question about the install process."
Do you have any good examples?
The one in the linked article doesn't rotate. You may mean https://www.theverge.com/2023/1/5/23541379/lenovo-thinkbook-... which will be available in summer and is pretty pricey.