General accepted definitions for open ML/AI are a must for Free and Open ML / AI systems and software. But despite several attempts there is still no general de facto accepted definition for ML/AI software and applications. The OSI Open Source AI Definition is not perfect. It is and was heavily criticized. However it is a first step towards a better definition.
1 Prompt Injections
Using LLMs with prompts is not without security risks. This is a nice short blog on visual prompt injection attack and how to recognize it.
(Link)
2 The Failed Strategy of Artificial Intelligence Doomers
New possibilities of LLMs are continuous threatened. Often by governments, scientist but often by people who are afraid of new technology and its impact on humans. This is an absolute must read article about the so called Foundations and coalitions that preach for safe or better AI. To avoid misunderstandings: I preach for open ML/AI. But I do not want to be in the camp of the ‘Doomers’.
(Link)
3 Demos featuring the latest AI research from Meta
I love the openness of Meta research and the products they developed with a FOSS license. But I dislike their shitty products that kill privacy and the lack of moral towards society. Playing with this demos is fun and gives nice insights.
(Link)
4 The Human Cost Of Our AI-Driven Future
The current progress on ML/AI and especially LLMs has a very high cost. Data workers in the most horrible places on earth do the hard and dirty work. When you preach for doing more with current LLM technology, you should read this horrible story. The news is not new but the misery is still going on outside our view.
(Link)
5 AI you can trust
Easily summarize emails, docs, articles, and videos across the web — without sacrificing your privacy. This sounds to good to be true. But its from Mozilla. A trusted party when it comes to Internet privacy. So check it out!
(Link)
6 The impact of Machine Learning models on the Web
AI systems have a systemic impact on the Web. But what about future Web Standards to steer the impact? This report is published by W3C, the consortium that develops standards and guidelines for the web. So an interesting read!
(Link)
7 TinyLlama: An Open-Source Small Language Model
FOSS AI/ML means doing more with less. So without large investments everyone can create a mind blowing LLM. This research work helps. Maybe DeepSeek used parts of this research?
(Link)
8 The Impact of Generative AI on Critical Thinking
Nice research paper of MS on the impact of Generative AI for knowledge workers and the impact on critical thinking skills and practices.
(Link)
9 AI is Killing How Developers Learn. Here’s How to Fix It
Foundational knowledge is crucial for making systems and software work. Programming is thinking. But AI tools do not simulate thinking. In the old days you could search for ‘copy-paste’ code from StackOverflow in code bases. Times change. But real professional system development for complex system is not yet possible with using LLMs. But it can help if you really know what you are doing.
(Link)
Our partners:
The FOSS Machine Learning News Blog is a brief overview of open machine learning news from all over the world. Free and Open machine learning means that everyone must be able to develop, test and play and deploy machine learning solutions. Read and share the FOSS ML Guide! And remember:You are invited to join the Free and Open Machine Learning open collaboration project.