Text-to-image diffusion models excel at generating high-quality, diverse images from natural language prompts. However, they often fail to produce semantically accurate results when the prompt contains concept combinations that contradict their learned priors. We define this failure mode as contextual contradiction, where one concept implicitly negates another due to entangled associations learned during training. To address this, we propose a stage-aware prompt decomposition framework that guides the denoising process using a sequence of proxy prompts. Each proxy prompt is constructed to match the semantic content expected to emerge at a specific stage of denoising, while ensuring contextual coherence. To construct these proxy prompts, we leverage a large language model (LLM) to analyze the target prompt, identify contradictions, and generate alternative expressions that preserve the original intent while resolving contextual conflicts. By aligning prompt information with the denoising progression, our method enables fine-grained semantic control and accurate image generation in the presence of contextual contradictions. Experiments across a variety of challenging prompts show substantial improvements in alignment to the textual prompt.

Language models demonstrate both quantitative improvement and new qualitative capabilities with increasing scale. Despite their potentially transformative impact, these new capabilities are as yet poorly characterized. In order to inform future research, prepare for disruptive new model capabilities, and ameliorate socially harmful effects, it is vital that we understand the present and near-future capabilities and limitations of language models. To address this challenge, we introduce the Beyond the Imitation Game benchmark (BIG-bench). BIG-bench currently consists of 204 tasks, contributed by 450 authors across 132 institutions. Task topics are diverse, drawing problems from linguistics, childhood development, math, common-sense reasoning, biology, physics, social bias, software development, and beyond. BIG-bench focuses on tasks that are believed to be beyond the capabilities of current language models. We evaluate the behavior of OpenAI's GPT models, Google-internal dense transformer architectures, and Switch-style sparse transformers on BIG-bench, across model sizes spanning millions to hundreds of billions of parameters. In addition, a team of human expert raters performed all tasks in order to provide a strong baseline. Findings include: model performance and calibration both improve with scale, but are poor in absolute terms (and when compared with rater performance); performance is remarkably similar across model classes, though with benefits from sparsity; tasks that improve gradually and predictably commonly involve a large knowledge or memorization component, whereas tasks that exhibit "breakthrough" behavior at a critical scale often involve multiple steps or components, or brittle metrics; social bias typically increases with scale in settings with ambiguous context, but this can be improved with prompting.

Researchers from Tel Aviv University have created a robot that can smell and identify odours using a biological sensor. The researchers connected the sensor to an electronic system. They used a machine learning algorithm to detect odours with a level of sensitivity that is 10,000 times higher than that of a commonly used electronic device. The sensor sends electrical signals as a response to the presence of a nearby odour, which the robot can detect and interpret. According to the University, the researchers say, "The sky's the limit," and believe this technology may also be used to identify explosives, drugs, diseases, and more.

I explain Artificial Intelligence terms and news to non-experts. This new model by Google Research and Tel-Aviv University is incredible. MyStyle is a very powerful deepfake model that can do basically anything. Take a hundred pictures of any person and you have its persona encoded to fix, edit or create any realistic picture you want. This is both amazing and scary, if you ask me, especially when you look at the results.

Whether teaching machine learning to undergrads, master students, or PhD students, I found myself time and time again choosing the 2012

A new technology developed at Tel Aviv University will make it possible, using artificial intelligence, to identify patients who are at risk of serious illness as a result of blood infections. The researchers trained the AI program to study the electronic medical records of about 8,000 patients at Tel Aviv's Ichilov Hospital who were found to be positive for blood infections. These records included demographic data, blood test results, medical history and diagnoses. After studying each patient's data and medical history, the program was able to automatically identify medical files' risk factors with an accuracy of 82%. According to the researchers, in the future this model could even serve as an early warning system for doctors, by enabling them to rank patients based on their risk of serious disease.

A new technology developed at Tel Aviv University will make it possible, using artificial intelligence, to identify patients who are at risk of serious illness as a result of blood infections. The researchers trained the AI program to study the electronic medical records of about 8,000 patients at Tel Aviv's Ichilov Hospital who were found to be positive for blood infections. These records included demographic data, blood test results, medical history and diagnosis. After studying each patient's data and medical history, the program was able to automatically identify medical files' risk factors with an accuracy of 82 percent. According to the researchers, in the future this model could even serve as an early warning system for doctors, by enabling them to rank patients based on their risk of serious disease.

A technology enabling the hearing-impaired to conduct phone conversations. Studies on the use of artificial intelligence (AI) to enhance disease diagnosis. These are just a few of the projects initiated by Google Israel intended to make the world a better place using artificial intelligence. Now those projects have a formal home: the "AI for Social Good" program established by Google and Tel Aviv University. The new three-year program, housed at the TAU Center for Artificial Intelligence and Data Science, launched in February under the direction of professor Meir Feder of TAU's Iby and Aladar Fleischman Faculty of Engineering.