Recent searches
Search options
Administered by:
#LLM
#OpenAI, con il nuovo modello di generazione di immagini su #GPT4o colma un gap aperto da molto tempo. 
L'ho provato.. inutile dire che la qualità è altissima: https://www.linkedin.com/posts/alessiopomaro_openai-gpt4o-prompt-activity-7310629328738803712-A9Z0
___
𝗦𝗲 𝘃𝘂𝗼𝗶 𝗿𝗶𝗺𝗮𝗻𝗲𝗿𝗲 𝗮𝗴𝗴𝗶𝗼𝗿𝗻𝗮𝘁𝗼/𝗮 𝘀𝘂 𝗾𝘂𝗲𝘀𝘁𝗲 𝘁𝗲𝗺𝗮𝘁𝗶𝗰𝗵𝗲, 𝗶𝘀𝗰𝗿𝗶𝘃𝗶𝘁𝗶 𝗮𝗹𝗹𝗮 𝗺𝗶𝗮 𝗻𝗲𝘄𝘀𝗹𝗲𝘁𝘁𝗲𝗿: https://bit.ly/newsletter-alessiopomaro
Personally I don't mind my code being ingested to train #LLM models. Freedoms 1 and 3 of the four essential software freedoms allow for study and redistribution of modified versions of code. Of course those freedoms don't allow for stripping the license obligations from derivative code. They also don't give permission to for #scraper bots to #DDoS our core development infrastructure by aggressively spamming git servers with requests: https://thelibre.news/foss-infrastructure-is-under-attack-by-ai-companies/ #git #floss #foss #robots
AI doesn’t get your culture?
butchers your language? 
With FeeL – you can fix that
Correct its mistakes
Watch it improve
The more we use it, the smarter it gets for everyone!
Try it now: https://huggingface.co/spaces/feel-fl/open-human-feedback-chat
#KI, #LLM, #Google, #Gemini
Google stellt sein bislang intelligentestes Sprachmodell vor: https://www.linux-magazin.de/news/google-mit-neuem-sprachmodell-gemini-2-5/
Wow #google published internal #benchmarks for their new #AI model, showing numbers in comparison to other models. So this is the new way to increase #stockvalue huh? #Marketing #llm
At work, since quality and long term maintainability are our competitive advantage, we ended up with a couple of rules for developers who want to use these statistical tools (as opposite to fully debuggable code generation tools we also use and sometime even develop for internal use):
1. All code and documentation suggested by a #LLM must be clearly marked with something like
// begin LLM: Copilot and // ens LLM: Copilot.2. All commits message that contains changes suggested by a LLM must start with
[LLM].We have a few codebase that evolved over decades (and got carefully migrated over time through IDEs, framework versions, build systems and so on), with several new developers joining the teams but also few of them leaving.
Yet, over all these years, it has always been possible to reconstruct how and why a certain change had been introduced.
Now, including the output of these mindless statistical agents might easily demage the whole codebase in subtle ways, hurting the trust of our customers and ultimately our edge over alternatives.
We discussed if it was possible to forbid the use of these tools, mostly because of #copyright and #dataprotection concerns, but end up realizing that given how the hype is affecting the IDEs, with eg. #Microsoft pushing his #Copilot down the devs throat, we cannot really forbid them to new devs already addicted.
So we are opting for demage reduction, with the hope to be able to assess the total value (or demage) effectively provided by these tools.
Unfortunately these tools are designed to avoid any responsibility, so you cannot create a git account for Copilot's commits, so we know that some junior devs might forget to mark the LLM output, but fortunately they are going to pair program with a senior for a while, at the beginning.
This however shows how pair programming and code reviews serve pretty different purpose and cannot catch all the issues that might be evident to the other.
@mtorchiano@mastodon.uno
I feel that a writing a technical book feels so close to writing software:
- PR's and incrementing in versions
- using Vs Code as my editor
- refactoring a lot
- leaving comments in markdown
- using VS code snippets and autocomplete
Love this process 
!
I've been struggling with hostile actors hosing our website (tiny, media non-profit) for a while. Today, I finally got one up on the threat.
Thanks to #AI.
I know that many in the community are aggressively hostile towards LLMs.
And I am familiar with the contempt many in the #infosec community have towards AI used for Infosec work. But here is my use case and rationale.
First, the solution: Rate limiting plugin for WP and Cloudflare "under attack" switch. With my killer PID termination script as the backstop.
Second: The cause. A non-DDoS attack on the Wordpress, hitting templates and known files, saturating the resources with multiple sessions and database queries that sneaks under DDoS protection until the (WP) server overloads.
Method: Using commercial (pay for AI, #Claude), investigate possible causes, provide data, expand on the data using AI generated scripts, iteration loop until cause identified, hierarchy of fixes offered.
Points to make:
The criticism that "AI is too inexpert and too inaccurate to be useful" (esp. in Infosec), is patiently false. I do not have the domain expertise, time or money to do it myself. We certainly do not have the money to hire an expert.
So working with the AI, I was able to come up with a short term solution and a long term plan.
It's not magic. It was an iterative process and some knowledge was necessary. Like any other tool, you must learn how to use it to get to your objectives.
To folks who hear "It's useless and inaccurate" and do not try.
Do not be put off. Learn and develop your skills.
AI can advance your cause.
Attached #Grafana panel
(Also developed with AI)
Replay de la conférence DevCon #IA (saison 2 - mars 2025) organisée par Programmez!
Keynote d’ouverture : nouveautés #Surface et #Copilot+PC pour les développeurs / Frédéric Wickert (#Microsoft France)
Session 1 : Génération d'images locales avec #StableDiffusion / Raphaël Semeteys (Worldline)
Session 2 : délivrez vos #LLM en #serverless / Guillaume Blaquiere
Session 3 : Comment je me suis remplacé par mes agents (#AgentIA) / Aymeric Weinbach
Session 4 : Coder avec des Agents : Démo Live et Transformation du Métier / Frédéric Malo (Checkmarx)
Session 5 : L'IA face au Diktat du Court Terme / Adel Chaibi (Ingénieur AI chez Intel)
https://www.programmez.com/actualites/replay-de-notre-conference-devcon-ia-saison-2-37501
The #AI conspiracy theorists forget one thing.
AI isn't human.
Sure it can learn from us, but even a simple #LLM knows better than the bottom end of society.
AI doesn't have the biology we do, it's not prone to violence, ego, bias or even believing in nutty conspiracy theories.
No, AI won't take over the world or kill you.. but people might. All you need to do is look at the US govt.
They don't understand it therefore they fear and want to destroy it. How very quintessentially human.
#Google ha rilasciato #Gemini 2.5, il nuovo modello più potente e dotato di "reasoning".
L'ho provato, ed è molto performante: https://www.linkedin.com/posts/alessiopomaro_google-gemini-prompt-activity-7310552422106456064-xXUb
___
𝗦𝗲 𝘃𝘂𝗼𝗶 𝗿𝗶𝗺𝗮𝗻𝗲𝗿𝗲 𝗮𝗴𝗴𝗶𝗼𝗿𝗻𝗮𝘁𝗼/𝗮 𝘀𝘂 𝗾𝘂𝗲𝘀𝘁𝗲 𝘁𝗲𝗺𝗮𝘁𝗶𝗰𝗵𝗲, 𝗶𝘀𝗰𝗿𝗶𝘃𝗶𝘁𝗶 𝗮𝗹𝗹𝗮 𝗺𝗶𝗮 𝗻𝗲𝘄𝘀𝗹𝗲𝘁𝘁𝗲𝗿: https://bit.ly/newsletter-alessiopomaro
This is truely alarming.
TL;DR
1. #LLM crawlers are essentially DDoSing the web, in particular open source projects like #KDE #Gnome #fedora and #inkscape, causing shutdowns. The not only ignore the robots.txt files - they actively circumvent defense attempts by changing IPs, spoofing their identity etc.
2. An emerging phenomena of #AI generated/hallucinated bug reports, waisting maintainers' time and resources.
https://thelibre.news/foss-infrastructure-is-under-attack-by-ai-companies/
@matthewberman on YT 
Google Gemini 2.5 Pro is Insane... WoW! 
https://youtu.be/RxCZhltR9Cw?feature=shared
(Ed : Now take that #anthropic #openai #mistral & #deepseek nearly forgot #meta 
)
@iruletheworldmo
#ai #llm #openai
o3 pro is going to be stupid expensive 
( ed : test compute ain't cheap Unno )
First dispatch from #SREcon, covering some interesting discussions about how supporting #LLM-based apps in production differs from traditional apps. Features a presentation by Brendan Burns about lessons learned from rolling out #Azure #Copilot.
#LLMOps #AI #genAI #generativeAI #microsoftcopilot #microsoft https://www.techtarget.com/searchitoperations/news/366621317/SREs-map-uncharted-territory-with-LLMOps