InfoQ Homepage Artificial Intelligence Content on InfoQ
-
The Code Written in Timefold Outsmarted the Solution Provided by ChatGPT for Devoxx Talks Scheduling
Stephan Jansen, Java Champion and Devoxx founder, tinkered with ChatGPT to “create the perfect conference schedule”. He managed to obtain feasible results for one day but not for the whole five-day conference. Geoffrey de Smet, founder of Timefold (formerly OptaPlanner), argued that the AI Assistant fails by not being able to treat soft constraints as effectively as a specialized planning library.
-
LLMs May Learn Deceptive Behavior and Act as Persistent Sleeper Agents
AI researchers at OpenAI competitor Anthropic trained proof-of-concept LLMs showing deceptive behavior triggered by specific hints in the prompts. Furthermore, they say, once deceptive behavior was trained into the model, there was no way to circumvent it using standard techniques.
-
Google Announces Video Generation LLM VideoPoet
Google Research recently published their work on VideoPoet, a large language model (LLM) that can generate video. VideoPoet was trained on 2 trillion tokens of text, audio, image, and video data, and in evaluations by human judges its output was preferred over that of other models.
-
Custom GPTs from OpenAI May Leak Sensitive Information
After it was reported that OpenAI has started rolling out its new GPT Store, it was also discovered that some of the data they’re built on is easily exposed. Multiple groups have begun finding that the system has the potential to leak otherwise sensitive information.
-
OpenAI GPT Store is a Nascent Marketplace for Custom ChatGPTs
OpenAI has started rolling out its new GPT Store, announced a few months ago along with GPTs, to provide a mechanism for ChatGPT Plus, Team and Enterprise users to share custom ChatGPT-based chatbots they create.
-
Researchers at Stanford Use Brain Signals to Control Intelligent Robots
In a paper presented at the 7th Annual Conference on Robot Learning last November, a team of Stanford University researchers presented an intelligent human brain-robot interface that enables controlling a robot through brain signals. Dubbed NOIR, short for Neural Signal Operated Intelligent Robots, the system uses electroencephalography (EEG) to communicate human intentions to the robots.
-
OpenAI Adopts Preparedness Framework for AI Safety
OpenAI recently published a beta version of their Preparedness Framework for mitigating AI risks. The framework lists four risk categories and definitions of risk levels for each, as well as defining OpenAI's safety governance procedures.
-
Griffin 2.0: Instacart Revamps Its Machine Learning Platform
Instacart created the next-generation platform based on experiences using the original Griffin machine-learning platform. The company wanted to improve user experience and help manage all ML workloads. The revamped platform leverages the latest developments in MLOps and introduces new capabilities for current and future applications.
-
Quarkus LangChain4J Extension Allows Developers to Integrate LLMs in Their Quarkus Applications
Inspired by the presentation “Java Meets AI” at Devoxx BE 2023, the Quarkus team started working on an extension based on the LangChain4J library, the Java re-implementation of the langchain library. This would allow developers to integrate LLMs Quarkus applications. The current is version, 0.5. The extension was built using Quarkus' usual declarative style, resembling the REST client.
-
Google Unveils Duet AI for Developers Offering AI-Powered Coding Assistance and Chat Functionality
Recently, Google announced the general availability (GA) of its Duet AI for Developers, providing developers with IDE coding assistant and chat.
-
OpenAI Publishes GPT Prompt Engineering Guide
OpenAI recently published a guide to Prompt Engineering. The guide lists six strategies for eliciting better responses from their GPT models, with a particular focus on examples for their latest version, GPT-4.
-
Using Gemini AI in Android Apps with the New Google AI SDK
Google introduced its new Google AI SDK to simplify integrating Gemini Pro, its best-performing model to date, in Android apps. Using this SDK, developers need not build and manage their own backend infrastructure.
-
Amazon Q Code Transformation: Automating Java Application Upgrades
AWS has recently announced the preview of Amazon Q Code Transformation, a service designed to simplify the process of upgrading existing Java application code through generative artificial intelligence. The new feature aims to minimize legacy code and automate common language upgrade tasks required to move off older language versions.
-
Microsoft Announces Small Language Model Phi-2
Microsoft Research announced Phi-2, a 2.7 billion-parameter Transformer-based language model. Phi-2 is trained on 1.4T tokens of synthetic data generated by GPT-3.5 and outperforms larger models on a variety of benchmarks.
-
Cloudflare's Journey in ML and AI: MLOps Platform and Best Practices
Cloudflare's blog described its MLOps platform and best practices for running Artificial Intelligence (AI) deployment at scale. Cloudflare's products, including WAF attack scoring, bot management, and global threat identification, rely on constantly evolving Machine Learning (ML) models. These models are pivotal in enhancing customer protection and augmenting support services.