Move fast, kill things: the tech startups trying to reinvent defence with Silicon Valley values
Venture capital-backed, $1bn companies are disrupting the way war will be waged with AI and futuristic weapons. Will they overthrow […]
2025
Meet the Fish That Doesn’t Want to Be Met
Welcome back to the Abstract! You’ve probably been reading a lot about humans this week. Most of the news seems
Meet the Fish That Doesn’t Want to Be Met Read Post »
Artificial neurons organize themselves
Novel artificial neurons learn independently and are more strongly modeled on their biological counterparts. A team of researchers has programmed
Artificial neurons organize themselves Read Post »
Cold plunges actually change your cells
Cold water immersion for seven days significantly improves cellular resilience and autophagic function, helping cells manage stress better. This adaptation
Cold plunges actually change your cells Read Post »
The CDC buried a measles forecast that stressed the need for vaccinations
ProPublica is a Pulitzer Prize-winning investigative newsroom. Sign up for The Big Story newsletter to receive stories like this one
The CDC buried a measles forecast that stressed the need for vaccinations Read Post »
First Look at Reasoning From Scratch: Chapter 1
Hi everyone, As you know, I’ve been writing a lot lately about the latest research on reasoning in LLMs. Before
First Look at Reasoning From Scratch: Chapter 1 Read Post »
Did Japan’s Copyright Loophole Spark the ‘Ghiblification’ of AI?
Grant Slatton, a founding engineer at Row Zero, used OpenAI’s 4o image generation to generate a Studio Ghibli version of
Did Japan’s Copyright Loophole Spark the ‘Ghiblification’ of AI? Read Post »
Hands on with Gemini 2.5 Pro: why it might be the most useful reasoning model yet
Gemini 2.5 Pro stands out with its massive context window, impressive multimodal reasoning and detailed reasoning chain.Read More
Hands on with Gemini 2.5 Pro: why it might be the most useful reasoning model yet Read Post »
Researchers warn of ‘catastrophic overtraining’ in LLMs
The researchers compared two versions of OLMo-1b: one pre-trained on 2.3 trillion tokens and another on 3 trillion tokens.Read More
Researchers warn of ‘catastrophic overtraining’ in LLMs Read Post »
New approach to agent reliability, AgentSpec, forces agents to follow rules
Researchers from Singapore Management University developed a new domain-specific language for agents to remain reliable.Read More
New approach to agent reliability, AgentSpec, forces agents to follow rules Read Post »