Author HubMatthias Bastian

Matthias is the co-founder and publisher of THE DECODER, exploring how AI is fundamentally changing the relationship between humans and computers.

AI research

Nov 11, 2025Nov 11, 2025

Update

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

AI in practice

Oct 31, 2025Oct 31, 2025

Google leans on token metrics, not revenue, adding to bubble talk about AI growth

AI and society

Oct 19, 2025

The long-predicted deepfake dystopia has arrived with Sora 2

AI in practice

Dec 12, 2025Dec 12, 2025

Short

Google has updated the voice for "Search Live." A new Gemini audio model powers the feature, producing responses that sound more natural and fluid, according to a blog post. Search Live lets users have real-time conversations while displaying relevant websites. The feature is part of Google Search's "AI Mode".

The update rolls out to all Search Live users in the US over the coming week. Users can open the Google app on Android or iOS, tap the Live icon, and speak their question.

The update fits into Google's broader push to build a voice-controlled assistant capable of handling everyday tasks—a goal shared by OpenAI and other major AI companies.

Google

AI in practice

Dec 12, 2025Dec 12, 2025

Short

According to Bloomberg's sources, Meta is shifting its focus to a new AI model codenamed "Avocado," with a release potentially coming next spring. Avocado is expected to launch as a closed model, letting the company sell access directly. This marks a major shift from Meta's established open-model strategy. Internally, the open-source approach reportedly lost steam after the disappointing performance of Llama 4. Management is betting big on Alexandr Wang, who joined Meta following the company's deal with Scale AI.

The development process involves some surprising ingredients. According to Bloomberg, the team is training Avocado using several external models, including Google's Gemma, OpenAI's gpt-oss, and Alibaba's Qwen. Using Chinese technology clashes with CEO Mark Zuckerberg's previous warnings about Chinese censorship.

Bloomberg

AI and society

Dec 10, 2025Dec 10, 2025

Author HubMatthias Bastian

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

Google leans on token metrics, not revenue, adding to bubble talk about AI growth

The long-predicted deepfake dystopia has arrived with Sora 2

Runway unveils first "General World Model" alongside major Gen-4.5 upgrades

Pangram 3.0 AI text detector claims up to 99.98% accuracy, even for subtly AI-assisted content

GPT-5.2 lands to top Google's Gemini 3 in the AI benchmark game just four weeks after GPT-5.1

Google opens updated Deep Research Agent to developers with new API

Google adds new features to boost website visibility in AI search

Deepseek reportedly using thousands of smuggled Nvidia chips for AI training

Mistral's open coding model Devstral 2 claims sevenfold cost advantage over Claude Sonnet

Big AI’s biggest names rally around the Agentic AI Foundation to set agent standards

GPT-5.2 lands to top Google's Gemini 3 in the AI benchmark game just four weeks after GPT-5.1

Corporate AI agents use simple workflows with human oversight instead of chasing full autonomy

Physicist Steve Hsu publishes research built around a core idea generated by GPT-5

Bank details