I'm not intrinsically opposed to generative models, even if I'm often voicing my criticism against how they're done, who uses them, and how they're used. Otherwise I wouldn't be subscribed to this comm. Plus I like the underlying tech on a conceptual level.
But there are limits on what those tools can do reliably, and often simpler tools would do the same job without, you know… cooking the planet. And that "reliably" is damn important, in some tasks 1% failure rate is already too much.
Someone might say "you can increase reliability with manual reviews". That's true but there's a catch: sometimes reviewing it takes longer than doing it yourself. Doubly so if it's a sensitive application, like the examples offered by the text:
Export your budget proposal to a Microsoft Excel (.xlsx) file
Now picture the model making some numbers up, and your proposal getting rejected because of that. Worse: picture someone claiming you've cooked the numbers. Gotta review it twice, thrice, five times.
Arrange loose ideas into a bulleted draft or consolidate a lengthy collaboration into a single-page PDF or Microsoft Word (.docx)
Until those collaborators get pissed at you, for distorting what they said. Remember, models don't summarise text to the meaning; they shorten it. Sometimes claiming the opposite of the original.
But you know, who this will help? People who don't care about bullshit. Now they can spread bullshit across multiple formats. "Yay".
How the AI bubble bursts
2mon 19d ago by sopuli.xyz/u/Fubarberry in gai@sopuli.xyz from martinvol.pe
Blocking the Internet Archive Won’t Stop AI, But It Will Erase the Web’s Historical Record
2mon 27d ago by lemmy.world/u/monica_b1998 in gai@sopuli.xyz from www.eff.org
The Chinese Core of “Uganda’s ChatGPT”
5mon 29d ago by lemmy.sdf.org/u/Hotznplotzn in gai@sopuli.xyz from chinamediaproject.orgGemini 3 deep think seems 👍
6mon 18h ago by lemmy.world/u/drawerair in gai@sopuli.xyz
Change the camera position of your input image - Qwen-Image-2509-MultipleAngles
7mon 11d ago by lemmy.ml/u/vermaterc in gai@sopuli.xyz from huggingface.co
Meet Project Suncatcher, Google’s plan to put AI data centers in space
7mon 14d ago by sopuli.xyz/u/Fubarberry in gai@sopuli.xyz from arstechnica.com
AI's Dial-Up Era
7mon 15d ago by sopuli.xyz/u/Fubarberry in gai@sopuli.xyz from www.wreflection.com
OpenAI data suggests 1 million users discuss suicide with ChatGPT weekly
7mon 21d ago by sopuli.xyz/u/Fubarberry in gai@sopuli.xyz from arstechnica.com
Alibaba’s AI bias problem: The Chinese tech giant’s language model reveals that English-language answers are more guided by the political leadership’s priorities than Chinese ones, a test finds
8mon 13d ago by lemmy.sdf.org/u/Hotznplotzn in gai@sopuli.xyz from chinamediaproject.org
Huawei co-develops DeepSeek model with stronger censorship tools
8mon 28d ago by lemmy.sdf.org/u/Hotznplotzn in gai@sopuli.xyz from www.techedt.com
How Generative AI Video Works | Computerphile [27:32]
9mon 14d ago by lemmy.sdf.org/u/ThorrJo in gai@sopuli.xyz from www.youtube.com
Google’s AI model just nailed the forecast for the strongest Atlantic storm this year
9mon 25d ago by sopuli.xyz/u/Fubarberry in gai@sopuli.xyz from arstechnica.com
Google just released Gemini 2.5 Flash Image, and it has some impressive image editing abilities
9mon 25d ago by sopuli.xyz/u/Fubarberry in gai@sopuli.xyz from developers.googleblog.com[Discussion] The future and direction of this community
9mon 25d ago by sopuli.xyz/u/Fubarberry in gai@sopuli.xyz
Being "Confidently Wrong" is holding AI back
9mon 29d ago by sopuli.xyz/u/Fubarberry in gai@sopuli.xyz from promptql.io
China's chatbots silent on Sichuan protests: Beijing-controlled AI models are now a crucial part of the Party’s censorship system for sudden-breaking stories and emergencies
10mon 5d ago by lemmy.sdf.org/u/Hotznplotzn in gai@sopuli.xyz from chinamediaproject.org
Introducing Gemma 3 270M: The compact model for hyper-efficient AI
10mon 6d ago by sopuli.xyz/u/Fubarberry in gai@sopuli.xyz from developers.googleblog.com
Why it’s a mistake to ask chatbots about their mistakes
10mon 9d ago by sopuli.xyz/u/Fubarberry in gai@sopuli.xyz from arstechnica.com
The GPT-5 rollout has been a big mess
10mon 9d ago by sopuli.xyz/u/Fubarberry in gai@sopuli.xyz from arstechnica.com
Claude Sonnet 4 Now Has a 1-million Token Context Window. Here's how it compares to Gemini's 1-million token context.
10mon 9d ago by sopuli.xyz/u/Fubarberry in gai@sopuli.xyz from every.to
OpenAI is releasing it's first open models in 6 years
10mon 16d ago by sopuli.xyz/u/Fubarberry in gai@sopuli.xyz from openai.com
Google Genie 3, for generating explorable worlds
10mon 16d ago by sopuli.xyz/u/Fubarberry in gai@sopuli.xyz from deepmind.google
Newelle, AI "Assistant" for GNOME, Hits Version 1.0
10mon 17d ago by sopuli.xyz/u/Fubarberry in gai@sopuli.xyz from www.omgubuntu.co.uk
All AI Models Might Be The Same
10mon 17d ago by sopuli.xyz/u/Fubarberry in gai@sopuli.xyz from blog.jxmo.io
LLM data privacy: Mistral least-privacy invasive platform, Meta AI, Google Gemini, Microsoft Copilot, Deepseek rank worst | Research
11mon 18d ago by lemmy.sdf.org/u/Hotznplotzn in gai@sopuli.xyz from blog.incogni.com
Top AI models - even American ones - parrot Chinese propaganda, report finds
11mon 25d ago by lemmy.sdf.org/u/Hotznplotzn in gai@sopuli.xyz from www.theregister.com
Why a US court allowed a dead man to deliver his own victim impact statement – via an AI avatar
11mon 29d ago by mander.xyz/u/Anyone in gai@sopuli.xyz from theconversation.com
Anthropic says most AI models, not just Claude, will resort to blackmail
11mon 1d ago by lemm.ee/u/neme in gai@sopuli.xyz from techcrunch.com
Characterizing censorship in DeepSeek: "AI-based censorship that subtly reshapes discourse rather than silencing it outright" | Research Report
11mon 2d ago by lemmy.sdf.org/u/Hotznplotzn in gai@sopuli.xyz from arxiv.org
Paper page - MoTE: Mixture of Ternary Experts for Memory-efficient Large Multimodal Models
11mon 2d ago by lemmy.world/u/cm0002 in gai@sopuli.xyz from huggingface.co
MiniMax M1 model claims Chinese LLM crown from DeepSeek – plus it's true open source
11mon 3d ago by lemmy.world/u/cm0002 in gai@sopuli.xyz from www.theregister.com
Unpacking the bias of large language models
11mon 3d ago by lemmy.world/u/cm0002 in gai@sopuli.xyz from news.mit.edu
Design Patterns for Securing LLM Agents against Prompt Injections
1y 9d ago by lemmy.world/u/cm0002 in gai@sopuli.xyz from simonwillison.net
This open-source generative AI tool converts 2D images into 3D part meshes for 3D printing in seconds
1y 9d ago by lemmy.world/u/cm0002 in gai@sopuli.xyz from www.tomshardware.comThe Quiet Voices Questioning China’s AI Hype: They’re easy to drown out, but worth listening to
1y 13d ago by lemmy.sdf.org/u/Hotznplotzn in gai@sopuli.xyz from www.bloomberg.com
China’s new DeepSeek censors even more than old models
1y 14d ago by lemmy.sdf.org/u/Hotznplotzn in gai@sopuli.xyz from www.newstarget.com
The first trial of generative AI therapy shows it might help with depression
1y 2mon ago by fedia.io/u/Fitik in gai@sopuli.xyz from www.technologyreview.comWhy I don’t like AI art – Cory Doctorow
1y 2mon ago by feddit.org/u/JiffyBag in gai@sopuli.xyz from pluralistic.net
AI Search Has A Citation Problem, Study Finds
1y 3mon ago by slrpnk.net/u/Anyone in gai@sopuli.xyz from www.cjr.org
can generative AI retell an Edgar Allan Poe story?
1y 3mon ago by lemm.ee/u/loomers in gai@sopuli.xyz from www.loomers.world
Australia bans DeepSeek on government devices over security risk
1y 4mon ago by lemmy.sdf.org/u/Hotznplotzn in gai@sopuli.xyz from www.bbc.com
No, DeepSeek isn’t uncensored if you run it locally
1y 4mon ago by lemmy.sdf.org/u/Hotznplotzn in gai@sopuli.xyz from techcrunch.com
The development of DeepSeek-V3 was probably much more expensive than suggested by the Chinese company, researchers say
1y 4mon ago by lemmy.sdf.org/u/Hotznplotzn in gai@sopuli.xyz from www.heise.deZuck's new Llama is a beast
1y 10mon ago by lemmy.world/u/drawerair in gai@sopuli.xyz
Marques Brownlee's latest vid is kinda unneeded
1y 11mon ago by lemmy.world/u/drawerair in gai@sopuli.xyz from y2u.beQuantized model issues
2y 7mon ago by sopuli.xyz/u/SpiderShoeCult in gai@sopuli.xyz