Ebryx, a cybersecurity leader, has launched LLMSec to support startups and mid-Market tech companies in safely adopting generative AI. As AI technologies like large language models grow, they bring new security challenges not addressed by traditional app security. LLMSec offers customized security services that protect AI systems against risks such as prompt injection, data leakage, ...
Large language models (LLMs) are transforming how we interact with technology, allowing users to communicate in natural language. In practical scenarios like scheduling appointments, AI agents can streamline processes by managing tasks such as checking calendars and verifying insurance all at once. However, challenges arise in managing these systems, such as optimizing tool selection and ...
A recent study has shown that the advanced AI model GPT-4.5 has successfully passed the Turing test, convincing people it’s human 73% of the time in a three-party setup. This new configuration tests an AI’s ability to imitate humans more effectively than before. The research involved participants engaging with both a human and an AI, ...
The journey from workflows to autonomous agents represents a significant shift in AI implementation. This evolution, driven by advanced language models, allows teams to create systems where AI can not only follow set processes but also adapt and improvise in real-time. Understanding the critical distinction between structured workflows and dynamic agents is crucial. While workflows ...
In this episode, we chat with Valentina Alto, a Technical Architect at Microsoft, about the exciting potential of AI Agents. Valentina explains why big companies are eager to adopt this technology, emphasizing how it can transform automation, boost productivity, and enhance customer experiences. She shares her expertise on the differences between AI Agents and traditional ...
Text-generating AI, like ChatGPT, is not just for answering questions; it can also advance fields like nuclear science. Research by Texas A&M graduate student Zavier Ndum shows how AI can automate tasks in this field, specifically through a project called AutoFLUKA. This tool enables users to run computer simulations, edit input files, and analyze results ...
Zavier Ndum, a nuclear engineering graduate student, is pioneering the use of text-generating AI, like ChatGPT, in nuclear science. His innovative project, AutoFLUKA, automates complex tasks such as running computer simulations and analyzing results securely, using proprietary data. By creating a domain-specific AI tool, Ndum addresses the challenge of confidentiality in nuclear research while enhancing ...
Recent advances in Large Language Models (LLMs) highlight the critical need to align their behavior with human intentions. This alignment challenge arises when an LLM’s goals differ from those of its developers, leading to potential risks, especially as these models gain more independence. While current methods help make LLMs safer, they may not adequately address ...
Recent advancements in Large Language Models (LLMs) have sparked concerns about their alignment with human goals. A significant issue is the potential for misalignment, where an LLM’s objectives may differ from those intended by its developers, posing serious risks. Current measures to align LLM behavior during training are helpful but may not be enough, especially ...
Behnam Mohammadi, a PhD student at the Tepper School of Business, focuses on human-AI interaction, large language models (LLMs), and AI regulation. His research includes developing LLM tools for businesses, like Pel, a programming language for coordinating AI agents, aimed at helping small businesses leverage AI. Mohammadi has published significant studies on AI transparency and ...
