AI

So apparently there's this online epidemic of "prophets" who are convinced their particular instance of ChatGPT or whatever other LLM they use is actually sentient, sapient, and why not - God.

There's debate about how normal people without explicit delusions get roped into this and effectively join an online cult.

But it's presented as if this happens by accident and that's what I don't buy. Your LLM is just going to parrot whatever you put in there. I've "asked" GPT if it's sentient as well and it answers with a firm "no" and then clarifies why. Although it does wax lyrical about it:

I'm curious where these grifters will lead people in the near future. We have plenty of examples from fiction, but in reality this is the first time we're seeing people worship technology and actually getting an answer.

I dunno man. I'm just a simple guy without high brow credentials or connections, but I can't see how people with actual critical thinking skills can possibly fall for this. Maybe the state of people's minds is just that much worse than even I in my pessimism and misanthropy imagine?___

Tony Stark speedrunning model collapse.

cross-posted from: https://lemmy.world/post/31763986

I started a webui container and then I started to get this error in OpenWebUI interface.

SyntaxError: JSON.parse: unexpected character at line 1 column 1 of the JSON data

• latest Ollama on windows
• latest Open WebUI in docker desktop

according to a post online, I should set ENABLE_WEBSOCKET_SUPPORT=True in my docker compose, but I'm not using reverse proxy. Is ENABLE_WEBSOCKET_SUPPORT=True necessary?

What could a possible solution be for this?

My docker compose

services:
  open-webui:
    image: ghcr.io/open-webui/open-webui:cuda 
    container_name: open-webui
    restart: unless-stopped
    ports:
      - "3000:8080"
    extra_hosts:
      - "host.docker.internal:host-gateway"
    volumes:
      - ./data:/app/backend/data
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: all
              capabilities: [gpu]
volumes:
  open-webui:

Title, or at least the inverse be encouraged. This has been talked about before, but with how bad things are getting, and how realistic goods ai generated videos are getting, anything feels better than nothing, AI generated watermarks, or metadata can be removed, but thats not the point, the point is deterrence. Immediately all big tech will comply (atleast on the surface for consumer-facing products), and then we will probably see a massive decrease in malicious use of it, people will bypass it, remove watermarks, fix metadata, but the situation should be quite a bit better? I dont see many downsides/

I asked teachers to tell me how AI has changed how they teach.

The response from teachers and university professors was overwhelming. In my entire career, I’ve rarely gotten so many email responses to a single article, and I have never gotten so many thoughtful and comprehensive responses.

One thing is clear: teachers are not OK.

They describe trying to grade “hybrid essays half written by students and half written by robots,” trying to teach Spanish to kids who don’t know the meaning of the words they’re trying to teach them in English, and students who use AI in the middle of conversation. They describe spending hours grading papers that took their students seconds to generate: “I've been thinking more and more about how much time I am almost certainly spending grading and writing feedback for papers that were not even written by the student,” one teacher told me. “That sure feels like bullshit.”

cross-posted from: https://lemmy.ml/post/30862067

Germany has been one of the worst Western countries for whitewashing Israel’s genocide in Palestine. Now it wants to do it with AI.

Becker was consulted by the Tagesspiegel because of his affiliation with the Decoding Antisemitism project at the Center for Research on Antisemitism at the Technical University Berlin, which he led from 2019 until 2025. With the help of a large language computing model, the project aims to create “an [AI] algorithm that will automatically recognize antisemitic statements in web comments . . . so that antisemitic posts can be removed more efficiently and accurately” by online platforms.

The dataset is divided into labels of differing forms of supposed antisemitisms such as “analogies with Nazism,” fascism, apartheid, or colonialism; calling Israel a racist or terrorist state; accusing it of genocide; referencing the Boycott, Divestment and Sanctions movement (BDS); giving Israel the sole blame for the plight of the Palestinians; applying double standards; and denying Israel’s right to exist.

On the topic of Palestine and Israel, the glossary seems to operate within a logic that sees emotional responses to a live-streamed genocide not as a human reaction but as an indicator of antisemitic beliefs.

In an interview with Israeli news outlet Mako, Becker suggests that social media providers are opening their doors and hearing concerns like his. This strongly suggests hopes to commercialize and implement its findings with online platforms. Five years after its inception, it appears that its conceptual framework and glossary have been overtaken by reality.

Asking out of curiosity

'Chicago Sun-Times' Slammed After Letting AI Generate Summer Reading List—Full Of Fake Book Titles

The spin doctors at it again. Waiting with glee for some toxic positive optimism. I live with and around the elderly. They want a sense of community so much they start believing in angels.

So, I have a 16GB vram GPU (4070 ti Super) and 32GB DDR4 RAM. The RAM is slow af and thus I tend to run models fully on GPU.

I can easily run up to 21b-ish models with Q4, sometimes high Q3.

I am testing various models out there but I was wondering if you guys have any reccommendation.

I am also really interested in understanding if quantization really decrease the model quality so much. Like, It would be better to have a Q6 12b model (like Gemma 3 12b), a Q2_K_L 32b model (such as QwQ 32b) or a Q3_XS model (such as Gemma 3 27b)?

Detailed manuscript critiques on all aspects of your novel — plot, prose, structure, and more. Powered by AI. Trained on editorial best practices.

Seems like an application that fits LLMs well.

For much of modern history, technological progress has been expected to lighten the burden of work. Keynes (1930) predicted that by 2030, rising productivity would allow people to work 15 hours a week. As artificial intelligence is integrated into workplaces, early evidence suggests a paradox: instead of reducing workloads, many AI-equipped employees are busier than ever. While AI-driven automation and delegation allow workers to complete the same tasks more efficiently, employees in AI-exposed occupations might well be working longer hours and spending less time on socialisation and leisure. …

• "Ein dog run dans le park, et son owner rennt après."
• "Un dog läuft in le parc, son master run après."
• "Ein doggy cour dans le parc, und der owner läuft nach ihm."

The silly language used in the 'Poland ball' memes got me wondering if that language was actually easier to understand for users of multiple languages. So I tried to get chatGPT to unite multiple languages (I tried Le Chat first but it wasn't as good).

My prompt:

Rewrite this phrase so that it is most understandable to people that only speak English, other people that only speak french, and a third group that only speaks German. You may use words from any language and you may misspell words if the word you use, although being mispelled, is more easily understood by speakers of both languages. You may not use slashes to separate two words for the same thing. You may change the phrase's structure as long as the meaning is the same. The phrase is: There is a dog running in the park. The owner is running after it

It's answer:

Here’s a version of the phrase rewritten to be easily understood by English, French, and German speakers:

"Un dog is courir in the park, und le master cour après."

Explanation of choices:

"Un dog": "Dog" is a simple English word that’s often understood by French and German speakers, and “un” is French for “a,” also recognizable to many.

"is courir": A mix of English “is” and French “courir” (to run), easy to guess even for German speakers who know “laufen” or “rennen.”

"in the park": Fully English, but all three languages share similar terms (“le parc,” “der Park”).

"und le master": “Und” is German for “and,” “le” is French for “the,” and “master” is close to “maître” (FR) and “Herrchen” (DE), but more internationally recognized.

"cour après": French phrase for “running after,” also intuitive to others.

I've used deep research on Gemini 2 and the results were absolutely incredible. To the point that I was shocked. There was no hallucination, no mistake and did what would take me 4 days of research and did it in under 2 mins. (I'm not saying it can't or won't make mistakes or hallucinations at other times for people, but this was clearly so much better than anything else I'd tried). And this wasn't even the best model. Its Gemini 2. Ive heard chat gpt has a research feature that is much better, although i haven't tried that. And according to industry folk, gpt 5 is another massive leap forward.

If you havent tried the deep research feature, you absolutely must.

cross-posted from: https://lemmy.world/post/28274434

(00:00)

Welcome to our first long format interview! Consider this a bonus episode. Please share it with others if you enjoy it! Let me know what you think; your feedback appreciated.

(00:20)

LinuxFest Northwest in Bellingham, WA April 25th - 27th

(00:37)

Quick Intro on Marcel - Developer behind Nextcloud Bookmarks, Floccus, Recognize

(01:04)

Recognize AI & ML for Nextcloud Photos documentation

• Project Github

(02:30)

Floccus - Browser Bookmark Syncing Extension for Chrome, Firefox, mobile clients, etc. Supports Nextcloud Bookmarks, Google Drive, Git, webdav and more.

• Project Github

(02:54)

Be sure to send in your feedback with this anonymous form!

(03:33)

Spread the word and share this show with others if you enjoy it! Thank you so much!

• You can donate to support me here.
• Podcasting 2.0 listeners to donate to support my upcoming Alby Hub node here. Fundraising 50k Sats.

(03:45)

Interview with Marcel Begins

• EfficientNet
• TensorFlow
• WhisperAI
• Stable Diffusion Image Generation by Stability AI
  • See some generated Mascots for Nextcloud
  • Try it here
  • Github repo

Beatles use AI to complete a new song

Nextcloud Assistant

• Project github
• Context Agent documentation

Summary Bot for Nextcloud Talk Chat

What are Common AI Models & How to Use Them

Ollama, supporting Deepseek and other kinds of models, from small to large.

• Project Github

Perplexica AI Search

• Built on Searxng

cross-posted from: https://lemmy.world/post/27865986

my go-to llms in no specific order –

1. Llama 4 behemoth
2. Grok 3
3. Claude 3.7 sonnet (extended thinking)

What's your go-to?

When in a hurry, I just use Gemini by telling my phone "Hey Google."

Washington (AFP) – Climate change deniers are pushing an AI-generated paper questioning human-induced warming, leading experts

Interesting discussion about using AI in science generally and Google AI co-scientist in particular.