Hub Artificial Intelligence

Artificial intelligence is either a field of research, the most important technical innovation of mankind, its downfall – or simply a pipe dream from Silicon Valley. Can machines really be intelligent? What is intelligence anyway? What opportunities does AI technology offer, and what are the risks? From neural networks to the science fiction vision of super AI, from deepfakes to AI surveillance: THE DECODER delivers the latest AI news and information on all facets of artificial intelligence.

Artificial Intelligence

Algorithmic prompting or how to teach math to a large language model

Artificial Intelligence

Meta Neuroscientist King: "Some of the concepts like reasoning may need to be re-evaluated"

Artificial Intelligence

DeepMind's Genie 2 generates playable 3D worlds from single images

Ad
Artificial Intelligence
Short

Cohere has introduced Embed 4, a multimodal language model designed for semantic search across complex enterprise documents. The model can process a wide range of content types—including text, images, tables, charts, code, and handwritten scans—commonly found in financial reports, medical records, and industrial documentation. Embed 4 supports files up to 128,000 tokens, or approximately 200 pages, and is compatible with over 100 languages, including Arabic, French, and Japanese. According to Cohere, the model is intended for organizations building language model-powered assistants that require access to internal knowledge. The model can be deployed either on-premises or in a private cloud environment, a configuration aimed at sectors with strict data sensitivity requirements, such as healthcare and manufacturing. Cohere says Embed 4 is now available through its own platform, as well as via Microsoft Azure AI Foundry and Amazon SageMaker.

Artificial Intelligence
Short

Apple is rolling out a new system that processes user data directly on devices in an effort to enhance its AI capabilities while maintaining privacy standards. Starting with beta versions of iOS and iPadOS 18.5 and macOS 15.5, the company will begin comparing synthetic data with real-world examples—such as emails from the Mail app—without storing this content or using it for model training. The approach is designed to address existing weaknesses in Apple’s AI systems without compromising user privacy. It builds on the company's emphasis on local processing and aligns with its broader privacy strategy. Improvements enabled by the system include more accurate notification summaries, better text input features, and expanded image-generation capabilities such as Genmoji. The system is only active for users who have opted into device analytics. The move comes as Apple works to close the gap with competitors like OpenAI and Google, both of which have made more visible progress in generative AI.

Ad
Ad
Ad
Ad
Artificial Intelligence
Short

The Irish Data Protection Commission (DPC) is investigating whether X used personal data from EU users without valid consent to train its AI system Grok. The investigation focuses on public posts from users in the EU and European Economic Area. X had previously committed to permanently stop this practice following a court case last year, which led the DPC to end its earlier investigation. The renewed scrutiny may have been triggered by Elon Musk's xAI acquiring X. As Ireland's lead EU regulator, the DPC can impose fines of up to four percent of a company's global revenue for violations of the General Data Protection Regulation. X was last fined €450,000 by the DPC in 2020. Elon Musk, X's owner, and former US President Donald Trump have repeatedly criticized EU regulations targeting US tech companies.

Ad
Ad
Google News

Bank details

IBAN: DE87 1203 0000 1086 0070 75
Account holder: DEEP CONTENT GbR
Purpose: Support THE DECODER
AI research

So-called reasoning models are more efficient but not more capable than regular LLMs, study finds

AI research

The next leap in AI depends on agents that learn by doing, not just by reading what humans wrote

AI in practice

OpenAI launches GPT-4.1: New model family to improve agents, long contexts and coding

Google News