Perplexity is a language-model evaluation metric derived from average cross-entropy loss. Intuitively, it reflects how many plausible choices the model appears to consider at each step; lower perplexity means the model assigns higher probability to the observed sequence. It is useful for comparing similar models on the same tokenized dataset, but it is not directly comparable across different vocabularies, tokenizers, or evaluation procedures. A lower score does not necessarily mean responses are more factual, helpful, safe, or preferred by people. Data contamination can also produce deceptively strong results. Perplexity is best treated as a training and predictive-fit measure alongside task-specific and human evaluations.
Barclays Research says humanoid robots are moving from experimental labs into real-world deployment as advances in AI and hardware reduce costs. The firm estimates the market could expand sharply over the next decade amid labor shortages.
U.S. senators are demanding detailed explanations from major tech platforms on how they prevent and monetize AI-generated sexual deepfakes. The inquiry follows renewed scrutiny of generative AI tools and their safeguards.
Microsoft outlined a community-first strategy for expanding AI data centers, promising to cover electricity costs and reduce local environmental impact. The move comes amid rising public opposition to large-scale AI infrastructure projects.
Apple introduced Apple Creator Studio, a subscription bundle combining professional creative apps with new AI-driven features across video, music, imaging, and productivity tools. The offering expands Apple’s push into intelligent software workflows for creators.
Meta has introduced a new top-level initiative, Meta Compute, to expand its AI infrastructure and energy capacity at large scale. The effort signals a long-term push to support advanced AI development and global data center growth.
Apple will rely on Google’s Gemini models and cloud infrastructure to power future artificial intelligence features, including Siri. The multi-year partnership supports Apple’s next generation of foundation models.
Lenovo introduced Qira, a system-level AI spanning PCs, tablets, smartphones, and wearables. The assistant offers context-aware, cross-device intelligence for writing, collaboration, and workflow management.
SoftBank has finalized its $40 billion investment in OpenAI, increasing its stake above 10%. The funding supports AI infrastructure initiatives, joint ventures, and OpenAI’s broader growth, including plans for a potential IPO.
The Association of Chartered Certified Accountants will discontinue remote exams from March 2026, citing rising misconduct linked to AI tools. Most candidates will return to in-person test centers.
OpenAI is hiring a Head of Preparedness to study emerging risks tied to rapidly advancing AI models, including mental health impacts and cybersecurity threats. The move reflects rising concern over how frontier capabilities could be misused as models grow more powerful.