Jon Seager, vice-president of Canonical for engineering and technical leader of the Ubuntu project, summarized plans for integrating functionality based on large language models into the distribution. In the first phase, Ubuntu developers intend to utilize AI models to enhance the existing functionality of the distribution, after which they will add separate AI capabilities and workflows for users interested in using AI. AI capabilities will be integrated into the distribution gradually throughout 2027, without any compulsion and as they are made ready.
When selecting AI tools, preference will be given to open models licensed in accordance with the spirit of the distribution, as well as to AI platforms based on open source. By default, AI models will run locally. Integration with external cloud AI systems will be available as an option controlled by the user.
It is emphasized that the goal of the initiative is not to turn Ubuntu into an AI product or to promote AI for the sake of AI, but to selectively integrate AI capabilities into the distribution where it will be genuinely beneficial for users. For those who do not wish to use AI, there will be an option to disable AI capabilities. Examples of AI applications include processing voice commands, diagnosing network and system issues, configuring services, analyzing logs, performing routine tasks, and conducting audits. servers.
The AI functionality added to the distribution is divided into implicit and explicit use of AI. In the first case, AI models expand existing features without changing the way users interact, for example, being used for speech recognition and synthesis. In the second case, AI models are offered as standalone solutions, such as AI agents for automating work, and AI assistants for learning, generating, and analyzing content.
Om de veiligheid te waarborgen en de installatie van AI-modellen te vereenvoudigen, is besloten om samen met de toolkits die voor hun uitvoering nodig zijn, snap-pakketten te leveren die zijn geoptimaliseerd voor verschillende hardwareplatforms. De modellen zullen worden uitgevoerd in een geĆÆsoleerde omgeving zonder directe toegang tot het overige systeem, terwijl AI-agenten zullen functioneren volgens de bestaande toegangsbeperkings- en auditmechanismen.
Binnen het bedrijf Canonical is besloten om in eerste instantie niet uit te gaan van een specifieke AI-stack, maar om de teams in de komende 6 maanden de mogelijkheid te bieden om zich te verdiepen in het onderwerp, verschillende AI-stacks uit te proberen, de optimale AI-oplossingen voor hun taken te kiezen en de sterke en zwakke punten van AI-tools te begrijpen. Medewerkers worden nog steeds beoordeeld op hoe goed ze hun taken uitvoeren, en niet op of ze AI gebruiken.
Bron: opennet.ru
