Modern language models (LLM) are transformer-based<\/strong> predictors that calculate the probability of the next token (word\/word fragment) given a context. The Transformer architecture replaced recurrent networks by using self-attention<\/strong>resulting in both better quality and significant training efficiency (Vaswani et al., 2017). arXiv+1<\/a><\/p>\n\n\n\n Scaling laws<\/strong> has steered the progress: DeepMind's \"Chinchilla\" showed that for compute-optimal<\/em> training should model size and number of training tokens are scaled approximately the same<\/strong>; many models were previously \"undertrained\" relative to their size. arXiv+1<\/a><\/p>\n\n\n\n In practice, 2025 models combine:<\/p>\n\n\n\n Limitations<\/strong> consists of: lack of lasting memory between sessions, uncertainty management, causal understanding of the world and robust planning outside of training allocation.<\/p>\n\n\n\n Artificial General Intelligence (AGI)<\/strong> is used for systems that can learn, reason and act broadly across domains-approaching human level or better. The very idea of \"machines that can outperform man\" is old: Turing discussed the test for machine \"thinking\" (1950), and I. J. Good outlined \"the first ultraintelligent machine\"<\/strong> who can improve themselves (1965\/66). courses.cs.umbc.edu+2incompleteideas.net+2<\/a><\/p>\n\n\n\n The term AGI<\/strong> in the modern sense was popularized in the early 2000s (Goertzel, Legg et al.), although some point to use dating back to the 1990s; the consensus is that AGI<\/em> was established as a terminology around 2002-2007 and consolidated through its own conferences and technical texts. Wikipedia+1<\/a><\/p>\n\n\n\n The difference from today's LLMs<\/strong> is not just about size: AGI requires architectures that connect language\/multimodal understanding with persistent memory, tool use, sensory and action loops<\/strong>, and goal-oriented learning<\/strong> and safe self-improvement - under management mechanisms that can be documented.<\/p>\n\n\n\n Major surveys of AI researchers (AI Impacts, 2023-2024) estimate approximately 50 % probability of HLMI\/AGI around 2047<\/strong>, with a very large spread; 10 % already in 2027 and many who believe significantly later. The results show both accelerated expectations from 2022 to 2023 and high disagreement. AI Impacts+2arXiv+2<\/a> Editorial review:<\/strong> Given the current method front (transform + scale + work with memory\/agentics) 2030s<\/strong> as the likely period for part-general<\/em> systems in delimited domains, while 2040s<\/strong> is a cautious middle ground for more broadly applicable, regulated \"AGI-like\" systems-conditional on security, regulation and infrastructure in place.<\/p>\n\n\n\n What they do:<\/strong> MediVox delivers AI-supported transcription and journal drafting for healthcare professionals with local language models operated in Sandefjord<\/strong> (stated by the company in previous communication with the editorial team and in public presentations). The solution addresses Norwegian clinical practice, documentation and compliance (GDPR\/ISO). Notice: For clinical claims and use cases, please refer to public notices\/contracts; health-related functionality should always be evaluated against WHO recommendations and the EU AI Act when use is \"high risk\". World Health Organization+1<\/a><\/p>\n<\/blockquote>\n\n\n\n What they do:<\/strong> EIR Tec describes a platform for home-based EEG<\/strong> and AI-supported analysis aimed at mental health\/neuro (e.g. ADHD\/ADD-related assessments). Official website: eirtech.co.uk. EIR TEC<\/a> What they do:<\/strong> ReadySOFT is building a .NET MAUI-based POS system (Android\/Sunmi) with AI modules for price optimization, menu\/range, campaign automation and predictive operations<\/strong>. Technical material and business pitch informs about local LLM infrastructure in ReadySOFT Cloud Sandefjord<\/strong>, Zero Data Retention<\/strong> using external LLM APIs, and integrations with Tripletex, PowerOffice and Fiken. <\/p>","protected":false},"excerpt":{"rendered":" Hva er AI\u2014og hvordan fungerer LLM-er i dag? Moderne spr\u00e5kmodeller (LLM) er transformer-baserte prediktorer som beregner sannsynligheten for neste token (ord\/ordfragment) gitt en kontekst. Transformer-arkitekturen erstattet rekurrente nettverk ved \u00e5 bruke self-attention, noe som ga b\u00e5de bedre kvalitet og kraftig oppl\u00e6ringseffektivitet (Vaswani mfl., 2017). arXiv+1 Skaleringslover har styrt fremgangen: DeepMinds \u201cChinchilla\u201d viste at for compute-optimal […]<\/p>","protected":false},"author":1,"featured_media":1634,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[42],"tags":[],"class_list":["post-1633","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-teknologi"],"_links":{"self":[{"href":"https:\/\/nettsak.no\/en\/wp-json\/wp\/v2\/posts\/1633","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/nettsak.no\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/nettsak.no\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/nettsak.no\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/nettsak.no\/en\/wp-json\/wp\/v2\/comments?post=1633"}],"version-history":[{"count":0,"href":"https:\/\/nettsak.no\/en\/wp-json\/wp\/v2\/posts\/1633\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/nettsak.no\/en\/wp-json\/wp\/v2\/media\/1634"}],"wp:attachment":[{"href":"https:\/\/nettsak.no\/en\/wp-json\/wp\/v2\/media?parent=1633"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/nettsak.no\/en\/wp-json\/wp\/v2\/categories?post=1633"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/nettsak.no\/en\/wp-json\/wp\/v2\/tags?post=1633"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}\n
What do we mean by AGI?<\/h2>\n\n\n\n
How can AI take the step towards AGI (technical building blocks)<\/h2>\n\n\n\n
\n
When can we expect AGI (timelines with uncertainty)<\/h2>\n\n\n\n
Popular science overviews and media coverage point to the same picture of uncertainty and divergent scientific camps. Our World in Data+1<\/a><\/p>\n\n\n\nWhere does the front stand internationally?<\/h2>\n\n\n\n
\n
\n\n\n\nWhat's happening in Norway (and near Norway) right now?<\/h1>\n\n\n\n
MediVox (Norway) - local models for clinical documentation<\/h2>\n\n\n\n
Who's behind it:<\/strong> Co-founded and led by Norwegian developers and clinicians; H\u00e5kon Berntsen is COO and co-founder.<\/p>\n\n\n\n\n
EIR Tec Ltd (UK) - EEG-based brain reading and home diagnostics<\/h2>\n\n\n\n
Who's behind it:<\/strong> Technology environment with Norwegian founding ties; H\u00e5kon Berntsen is CTO\/co-founder (disclosed to the editors).
Editorial clarification:<\/strong> Diagnostics<\/strong> in a legal\/medical sense requires regulatory approvals (class, indication, market). EIR Tec communicates ambitions for diagnostic support; readers should distinguish between clinical support tool<\/em> and formal diagnosis<\/em>.<\/p>\n\n\n\nReadySOFT \/ ReadyPOS (Norway) - AI-powered POS for SMBs<\/h2>\n\n\n\n
Who's behind it:<\/strong> The team includes Per Joar Lorvik (CEO)<\/strong>, \u00d8yvind Horn (CFO)<\/strong>, Helge Andresen (CTO\/Advisor)<\/strong>, and H\u00e5kon Berntsen (AI\/architecture, advisor)<\/strong>(Source: investor\/product documents submitted to the editors).<\/p>\n\n\n\n
\n\n\n\nIn short (editorial)<\/h2>\n\n\n\n
\n
\n\n\n\nSources (selection)<\/h2>\n\n\n\n
\n