Articles tagged with Natural Language Generation
Researchers at PolyU developed an AI-driven assessment system, LMLPA, to quantify LLM personality traits. The system analyzes linguistic patterns and style in LLM outputs to evaluate their personalities, with applications in education, manufacturing, business, and sustainable development.
A recent study from UTSA researchers reveals that large language models (LLMs) can pose a serious threat to programmers who use them to help write code. The study found that up to 97% of software developers incorporate generative AI into their workflow, and 30% of code written today is AI-generated.
Researchers have developed a hybrid image-generation tool called HART that combines the strengths of autoregressive and diffusion models. It achieves high reconstruction quality with significantly reduced computational resources, enabling local execution on laptops or smartphones.
A new AI-based tool can translate a person's thoughts into continuous text without requiring language comprehension, and it can be trained in under an hour. The system was developed by adapting a previous brain decoder to a new person using short, silent videos.
BioChatter bridges the gap between large language models and biomedical research by providing a transparent and adaptable framework for custom research tasks. The platform can integrate with knowledge graphs and bioinformatics tools, making it easier for researchers to analyze complex datasets.
Researchers introduced a novel approach to enhance reservoir computing, incorporating a generalized readout that offers improved accuracy and robustness compared to conventional methods. The new method uses a nonlinear combination of reservoir variables to uncover deeper patterns in input data.
A recent study by ESMT Berlin scholars presents a comprehensive framework detailing the impact of AI on open innovation. The framework highlights three key ways in which AI is transforming open innovation practices: enhancing existing methods through efficiency and scalability, enabling new forms of collaboration and business models, a...
A new system, EXPLINGO, enables AI models to generate readable narratives explaining their predictions, helping users make better decisions. The system, developed by MIT researchers, uses large language models to transform complex explanations into plain language and automatically evaluate their quality.
Researchers at University of Bath and Technical University of Darmstadt found that large language models like ChatGPT cannot learn independently or acquire new skills, making them controllable and predictable. The study concluded that LLMs remain inherently safe, but misuse is still possible.
Researchers investigated the efficiency of modern neural network-based generative models, comparing them to traditional sampling techniques. The study found that modern diffusion-based methods may face challenges due to a first-order phase transition, but also exhibit superior efficiency in certain cases.
Researchers from MIT develop a new technique called Natural Language Embedded Programs (NLEPs) to enable large language models to solve numerical, analytical, and language-based tasks. NLEPs achieved greater than 90 percent accuracy on symbolic reasoning tasks and showed 30 percent greater accuracy than task-specific prompting methods.
Researchers from MIT and the MIT-IBM Watson AI Lab devised a navigation method that converts visual representations into pieces of language, which are then fed into one large language model. The approach utilizes purely language-based representations, generating synthetic training data to overcome challenges in visual data availability.
Six UTA faculty members have received prestigious CAREER grants from the National Science Foundation for their innovative research. The total award amount is $3.23 million.
Nach0 was trained on diverse tasks, including natural language understanding, synthetic route prediction, and molecular generation. The model performed well on molecular tasks using molecular data and outperformed ChatGPT, making it a significant step toward unlocking the full potential of LLMs for drug discovery.
Researchers found that large language model-based chatbots fail to recognize users' motivational states when they are hesitant about making healthy behavior changes. However, the chatbots can provide relevant information and support users who have established goals and a commitment to take action. The study highlights the limitations o...
A study by Washington State University found ChatGPT's generative AI system provided inconsistent heart risk assessments for patients with chest pain. The AI failed to match traditional methods used by physicians and returned different results for the same patient data, highlighting its limitations in high-stakes clinical situations.
Researchers used generative AI to design a lead molecule for treating fibrosis, a biological process associated with aging. The compound, INS018_055, demonstrated significant efficacy in preclinical studies and showed promising results in clinical trials, accelerating drug discovery and providing new therapeutic options.
A new study found that high-peace countries are characterized by an increased prevalence of words related to optimism for the future and fun, while low-peace countries feature more references to control and fear. The research used a machine learning model to identify these linguistic patterns in media articles from 18 countries.
Researchers at MIT found that similarity-focused generative AI models falter when tasked with designing new products, highlighting the need to prioritize innovation in engineering tasks. By adjusting training objectives and metrics, AI can be an effective 'co-pilot' for engineers, enabling faster creation of innovative products.
Researchers developed a novel approach to predict therapeutic targets for aging and age-related diseases. They trained a domain-specific BioGPT model on biomedical literature, which improved its performance in identifying prospective targets.
Researchers used large language models to generate open-ended answers, which were often more convincing than real responses. This approach may help gather data quickly and at low cost but also raises concerns about the authenticity of online user data.
Researchers found that using positive trigger words can retrain large language models and result in less biased responses. The team analyzed GPT-2's responses to user prompts about different countries worldwide and found a significant impact on the types of adjectives used to describe citizens.
A recent study by Cornell University researchers found that lawmakers are only slightly more likely to respond to AI-generated messages than human-generated ones, highlighting the potential risks of emerging technologies on democratic representation. The research used a field experiment to investigate the impact of natural language mod...
A Penn State-led research team found that language models can plagiarize content in three ways: verbatim, paraphrase, and idea reuse. The study highlights the need for more research into text generators and their potential ethical implications.
A research team at Carnegie Mellon University is working to simplify data requirements for speech recognition models, aiming to reach 2,000 languages. By focusing on linguistic elements common across many languages and using a phylogenetic tree, the team hopes to eliminate the need for audio data.