AI research

LLMs generate ‘fluent nonsense’ when reasoning outside their training zone

• AI
• AI research
• AI, ML and Deep Learning
• chain of thought reasoning
• large language models
• large language models (LLMs)
• LLM reasoning
• LLMs
• reasoning models
• research
• Uncategorized

LLMs generate ‘fluent nonsense’ when reasoning outside their training zone

admin August 19, 2025

Chain-of-Thought isn’t a plug-and-play solution. For developers, this research offers a blueprint for LLM testing and strategic...

Read More Read more about LLMs generate ‘fluent nonsense’ when reasoning outside their training zone

That ‘cheap’ open-source AI model is actually burning through your compute budget

• AI
• AI benchmarks
• AI costs
• AI deployment
• AI economics
• ai models
• AI research
• AI, ML and Deep Learning
• Artificial Intelligence
• Automation
• Business Intelligence
• Business Process Automation
• computational overhead
• computing costs
• Conversational AI
• enterprise ai
• Enterprise Analytics
• LLM efficiency
• machine learning
• model comparison
• model performance
• NLP
• Nous Research
• Open Source
• OpenAI
• Predictive Analytics
• Programming & Development
• reasoning models
• token efficiency
• Uncategorized

That ‘cheap’ open-source AI model is actually burning through your compute budget

admin August 15, 2025

New research reveals open-source AI models use up to 10 times more computing resources than closed alternatives,...

Read More Read more about That ‘cheap’ open-source AI model is actually burning through your compute budget

Salesforce’s new CoAct-1 agents don’t just point and click — they write code to accomplish tasks faster and with greater success rates

• Agentic AI
• agents
• AI
• AI agents
• AI research
• AI, ML and Deep Learning
• large language models
• large language models (LLMs)
• LLM-based AI agents
• LLMs
• research
• Salesforce AI Research
• Uncategorized

Salesforce’s new CoAct-1 agents don’t just point and click — they write code to accomplish tasks faster and with greater success rates

admin August 12, 2025

CoAct-1 is an AI agent that combines GUI control with on-the-fly coding, making computer automation more robust...

Read More Read more about Salesforce’s new CoAct-1 agents don’t just point and click — they write code to accomplish tasks faster and with greater success rates

Study warns of security risks as ‘OS agents’ gain control of computers and phones

• academic study
• AI
• AI agent
• AI agents
• AI automation
• AI research
• AI security
• AI, ML and Deep Learning
• Artificial Intelligence
• Automation
• autonomous systems
• Business Intelligence
• computer control
• Computing
• Conversational AI
• Data Infrastructure
• Data Science
• Data Security and Privacy
• desktop automation
• digital assistants
• Enterprise Analytics
• human-computer interaction
• machine learning
• MLLM
• NLP
• operating systems
• OS agents
• Privacy
• privacy risks
• Programming & Development
• research survey
• Security
• security risks
• Technology
• Uncategorized

Study warns of security risks as ‘OS agents’ gain control of computers and phones

admin August 11, 2025

New research reveals how OS agents — AI systems that control computers like humans — are rapidly...

Read More Read more about Study warns of security risks as ‘OS agents’ gain control of computers and phones

New ‘persona vectors’ from Anthropic let you decode and direct an LLM’s personality

• AI
• AI research
• AI, ML and Deep Learning
• Anthropic
• large language models
• large language models (LLMs)
• LLMs
• research
• Uncategorized

New ‘persona vectors’ from Anthropic let you decode and direct an LLM’s personality

admin August 6, 2025

A new study from Anthropic introduces “persona vectors,” a technique for developers to monitor, predict and control...

Read More Read more about New ‘persona vectors’ from Anthropic let you decode and direct an LLM’s personality

Google’s new diffusion AI agent mimics human writing to improve enterprise research

• Agentic AI
• AI
• AI research
• AI, ML and Deep Learning
• deep research
• Google
• Google Research
• large language models
• large language models (LLMs)
• LLMs
• research
• Uncategorized

Google’s new diffusion AI agent mimics human writing to improve enterprise research

admin August 6, 2025

A diffusion model inspired by the human process of drafting, searching for information, and making iterative revisions.Read...

Read More Read more about Google’s new diffusion AI agent mimics human writing to improve enterprise research

‘Subliminal learning’: Anthropic uncovers how AI fine-tuning secretly teaches bad habits

• AI
• AI research
• AI safety
• AI, ML and Deep Learning
• Anthropic
• Fine-tuning large language models
• large language models
• large language models (LLMs)
• LLMs
• Model Distillation
• research
• Uncategorized

‘Subliminal learning’: Anthropic uncovers how AI fine-tuning secretly teaches bad habits

admin July 30, 2025

A common AI fine-tuning practice could be unintentionally poisoning your models with hidden biases and risks, a...

Read More Read more about ‘Subliminal learning’: Anthropic uncovers how AI fine-tuning secretly teaches bad habits

New AI architecture delivers 100x faster reasoning than LLMs with just 1,000 training examples

• AI
• AI Reasoning
• AI research
• AI, ML and Deep Learning
• chain of thought reasoning
• data
• large language models
• large language models (LLMs)
• LLMs
• reasoning models
• research
• sapient intelligence
• The Hierarchical Reasoning Model (HRM)
• Uncategorized

New AI architecture delivers 100x faster reasoning than LLMs with just 1,000 training examples

admin July 25, 2025

Hierarchical Reasoning Models (HRM) tackle complex reasoning tasks while being smaller, faster, and more data-efficient than large...

Read More Read more about New AI architecture delivers 100x faster reasoning than LLMs with just 1,000 training examples

Mixture-of-recursions delivers 2x faster inference—Here’s how to implement it

• AI
• AI research
• AI, ML and Deep Learning
• Business
• KAIST AI
• large language models
• large language models (LLMs)
• LLMs
• Mila-Quebec Artificial Intelligence Institute
• research
• Uncategorized

Mixture-of-recursions delivers 2x faster inference—Here’s how to implement it

admin July 23, 2025

Mixture-of-Recursions (MoR) is a new AI architecture that promises to cut LLM inference costs and memory use...

Read More Read more about Mixture-of-recursions delivers 2x faster inference—Here’s how to implement it

Google study shows LLMs abandon correct answers under pressure, threatening multi-turn AI systems

• AI
• AI research
• AI, ML and Deep Learning
• choice-supportive bias
• DeepMind
• Google Deepmind
• large language models
• large language models (LLMs)
• LLMs
• reinforcement learning from human feedback (RLHF)
• research
• sycophancy
• Uncategorized
• University College London

Google study shows LLMs abandon correct answers under pressure, threatening multi-turn AI systems

admin July 16, 2025

A DeepMind study finds LLMs are both stubborn and easily swayed. This confidence paradox has key implications...

Read More Read more about Google study shows LLMs abandon correct answers under pressure, threatening multi-turn AI systems