Google w 2025 roku – sztuczna inteligencja staje się podstawą produktów

Czy 2025 był rokiem, w którym Google naprawdę „włączyło” sztuczną inteligencję? Nie jako gadżet do podziwiania na keynote’ach, tylko jako działającą infrastrukturę pod produktami, pracą i – ambitnie – misjami publicznymi.

Czytaj dalej:
https://pressmind.org/google-w-2025-roku-sztuczna-inteligencja-staje-sie-podstawa-produktow/

#PressMindLabs #agenciai #gemini3 #gemma3n #google #medgemma

Mô hình lập trình vibe đang hot tháng 10/2025:
- Google Gemini 2.5 Pro cho trải nghiệm tích cực.
- Qwen Code, Ollama (cập nhật 5 tháng trước), Germma3n là những cái tên đáng chú ý.
- Cá nhân dùng Macbook Pro M5, RAM 28GB; Intel Core Ultra 7 265k, 5070 TI 16GB.
Trước khi dùng gói trả phí của Claude Code/GPT Code, hãy thử trên máy mình.
#LapTrinhVibe #CodingModel #GoogleGemini #QwenCode #Ollama #Gemma3n #MacbookPro #IntelCoreUltra7
🔥🖥️🌐

https://www.reddit.com/r/LocalLLaMA/comments/1ojjxvd/wh

Mô hình AI Gemma 3n hoạt động hiệu quả trên chip di động Snapdragon 6 Gen 1, chứng tỏ khả năng của chip di động trong việc chạy các mô hình AI cục bộ. Thậm chí vượt trội hơn một số laptop tích hợp card đồ họa.
#Gemma3n #Snapdragon6Gen1 #AI #LocalAI #MobileAI #ChipDiĐộng #MôHìnhAI #TríTuệNhânTạo

https://www.reddit.com/r/LocalLLaMA/comments/1o54vs1/gemma_3n_on_snapdragon_6_gen_1_processor/

In case anyone wants to run their own #LLM on their personal devices, I can suggest #lmstudio It allows you to download any major model that's out there to your own laptop/desktop, and then you can use it's Graphical Interface to interact with it in a familiar interface environment, as you would #Claude or #ChatGPT or #Gemini. It will also let you know if your device has enough resources (power) to run a specific model variant: https://lmstudio.ai/

For those of you who love the Terminal (aka: Command Line Interface or #CLI), check out https://ollama.com/

For starters, I suggest the #gemma3n model (works great on tablets, laptops, or phones), or #llama3.1 for the most common interactions. While most of us geeks will have very powerful personal systems, or servers in our home racks, most people do not have high-end systems/devices. The gemma3n model is lightweight, very powerful, and a solid general purpose LLM.

What's the benefit of running LLM's locally? #Privacy is a big one (it's running on your local machine, not a cloud server) - so you can ask it questions against sensitive business data, PHI/PII, etc.. You can also run it #offline (no Internet connection required) so if you wanted to #airgap your interactions, or play with it on vacation (on a plane, lost in the back country, etc.), you can absolutely do that - even with the Deepseek model.

I started playing with both LM Studio and ollama myself - been asking models basic questions like "what is the capital of Italy" and also more complex questions like "write me a #powershell script to add users of a specific OU to a number of Security Groups within AD" and so far, it's been very accurate. The PowerShell script llama3.1 provided worked out of the box (after I revised variables to match my environment).

#AIForward #AI #AiResearch #FunWithAI

🧠 #Gemma3n di Google in azione su task multimodali in locale sul mio smartphone.
✨ Negli esempi si vede come il modello interpreta il contenuto delle immagini, estraendo le informazioni richieste.
💡 Sembra una banalità, ma l'aspetto straordinario è che il #LLM sta funzionando in locale su un dispositivo con un hardware limitatissimo.

___
✉️ 𝗦𝗲 𝘃𝘂𝗼𝗶 𝗿𝗶𝗺𝗮𝗻𝗲𝗿𝗲 𝗮𝗴𝗴𝗶𝗼𝗿𝗻𝗮𝘁𝗼/𝗮 𝘀𝘂 𝗾𝘂𝗲𝘀𝘁𝗲 𝘁𝗲𝗺𝗮𝘁𝗶𝗰𝗵𝗲, 𝗶𝘀𝗰𝗿𝗶𝘃𝗶𝘁𝗶 𝗮𝗹𝗹𝗮 𝗺𝗶𝗮 𝗻𝗲𝘄𝘀𝗹𝗲𝘁𝘁𝗲𝗿: https://bit.ly/newsletter-alessiopomaro

#AI #IntelligenzaArtificiale #LLM

So I joked with the new #Gemma3n about visiting Sable Island (a lovely science-focused island with wild horses off the coast of eastern Canada) by taking a boat and sneaking in during cloak of night.

It recognized this as a crime, and dressed me down for it. But then it dumps Suicide & Crisis Lifeline and "LGBTQ youth" emergency numbers at me. (As it often does, since it's some baked-in safety nonsense.)

So I called it out, saying it's weird to hit me with those when all I'm doing is trespassing. It's usual creepy "you're right to call me out!" subservience is what I got in return... along with another blast with the help line numbers.

So I yelled at it again, and it verbally cowers and says I'm right, and woe is me -- and then hits me with the numbers again. We go 4-5 times through this cycle of "oh god why did I do that" and doing it again in the same response.

This time, I said "you've got a real problem", and it AGREES and tries to tell me what's going on from it's perspective.

None of this is surprising, but I'm still GREATLY amused even after all these years, that we can have conversations like this. 😏

"You're right! I'm horrible! I'm a shitty AI! Punish me, daddy!"

#llm #ollama #ai

🧠 Ho provato #Gemma3n in locale, su un laptop attraverso #Ollama e anche su #Colab. 

👉 Com'è andata? https://www.linkedin.com/posts/alessiopomaro_gemma3n-ollama-colab-activity-7345330146146410496-0B1T

___ 

✉️ 𝗦𝗲 𝘃𝘂𝗼𝗶 𝗿𝗶𝗺𝗮𝗻𝗲𝗿𝗲 𝗮𝗴𝗴𝗶𝗼𝗿𝗻𝗮𝘁𝗼/𝗮 𝘀𝘂 𝗾𝘂𝗲𝘀𝘁𝗲 𝘁𝗲𝗺𝗮𝘁𝗶𝗰𝗵𝗲, 𝗶𝘀𝗰𝗿𝗶𝘃𝗶𝘁𝗶 𝗮𝗹𝗹𝗮 𝗺𝗶𝗮 𝗻𝗲𝘄𝘀𝗹𝗲𝘁𝘁𝗲𝗿: https://bit.ly/newsletter-alessiopomaro 

#AI #GenAI #GenerativeAI #IntelligenzaArtificiale #LLM 

"Extremely consequential new open weights model release from Google today:
(...)
This is very exciting: a 2B and 4B model optimized for end-user devices which accepts text, images and audio as inputs!

Gemma 3n is also the most comprehensive day one launch I've seen for any model: Google partnered with "AMD, Axolotl, Docker, Hugging Face, llama.cpp, LMStudio, MLX, NVIDIA, Ollama, RedHat, SGLang, Unsloth, and vLLM" so there are dozens of ways to try this out right now.

So far I've run two variants on my Mac laptop."

https://simonwillison.net/2025/Jun/26/gemma-3n/

#AI #GenerativeAI #Google #LLMs #Chatbots #Gemma #Gemma3n