Terence Tao, a mathematics professor at UCLA, is a real-life superintelligence. The "Mozart of Math," as he is sometimes called, is widely considered the world's greatest living mathematician. He has won numerous awards, including the equivalent of a Nobel Prize for mathematics, for his advances and proofs. Right now, AI is nowhere close to his level. But technology companies are trying to get it there.

OpenAI announced this week that it has raised 6.6 billion in new funding and that the company is now valued at 157 billion overall. This is quite a feat for an organization that reportedly burns through 7 billion a year--far more cash than it brings in--but it makes sense when you realize that OpenAI's primary product isn't technology. Case in point: Last week, CEO Sam Altman published an online manifesto titled "The Intelligence Age." In it, he declares that the AI revolution is on the verge of unleashing boundless prosperity and radically improving human life. "We'll soon be able to work with AI that helps us accomplish much more than we ever could without AI," he writes.

The phrase is so associated with Caro that it's the name of the recent documentary about him and of an exhibit of his archives at the New York Historical Society. To Caro it is imperative to put eyes on every line of every document relating to his subject, no matter how mind-numbing or inconvenient. He has learned that something that seems trivial can unlock a whole new understanding of an event, provide a path to an unknown source, or unravel a mystery of who was responsible for a crisis or an accomplishment. Over his career he has pored over literally millions of pages of documents: reports, transcripts, articles, legal briefs, letters (45 million in the LBJ Presidential Library alone!). Some seemed deadly dull, repetitive, or irrelevant.

Meta just announced its own media-focused AI model, called Movie Gen, that can be used to generate realistic video and audioclips. The company shared multiple 10-second clips generated with Movie Gen, including a Moo Deng-esque baby hippo swimming around, to demonstrate its capabilities. While the tool is not yet available for use, this Movie Gen announcement comes shortly after its Meta Connect event, which showcased new and refreshed hardware and the latest version of its large language model, Llama 3.2. Going beyond the generation of straightforward text-to-video clips, the Movie Gen model can make targeted edits to an existing clip, like adding an object into someone's hands or changing the appearance of a surface. In one of the example videos from Meta, a woman wearing a VR headset was transformed to look like she was wearing steampunk binoculars.

Google's new AI podcasting tool, called Audio Overview, has become a surprise viral hit. The podcasting feature was launched in mid-September as part of NotebookLM, a year-old AI-powered research assistant. NotebookLM, which is powered by Google's Gemini 1.5 model, allows people to upload content such as links, videos, PDFs, and text. They can then ask the system questions about the content, and it offers short summaries. The tool generates a podcast called Deep Dive, which features a male and a female voice discussing whatever you uploaded.

Fear of missing out has rocketed the value of artificial intelligence companies, despite few signs as to when the technology will turn a profit, raising talk of AI overenthusiasm. The mystery deepens when it comes to predicting which generative AI firms will prevail, according to analysts. ChatGPT-maker OpenAI secured 6.6 billion in a funding round that propelled its valuation to an eye-popping 157 billion, sparking new worries there is an AI bubble poised to burst.

The process of satisfying daily demands is a fundamental aspect of humans' daily lives. With the advancement of embodied AI, robots are increasingly capable of satisfying human demands. Demand-driven navigation (DDN) is a task in which an agent must locate an object to satisfy a specified demand instruction, such as ``I am thirsty.'' The previous study typically assumes that each demand instruction requires only one object to be fulfilled and does not consider individual preferences. However, the realistic human demand may involve multiple objects. In this paper, we introduce the Multi-object Demand-driven Navigation (MO-DDN) benchmark, which addresses these nuanced aspects, including multi-object search and personal preferences, thus making the MO-DDN task more reflective of real-life scenarios compared to DDN. Building upon previous work, we employ the concept of ``attribute'' to tackle this new task. However, instead of solely relying on attribute features in an end-to-end manner like DDN, we propose a modular method that involves constructing a coarse-to-fine attribute-based exploration agent (C2FAgent). Our experimental results illustrate that this coarse-to-fine exploration strategy capitalizes on the advantages of attributes at various decision-making levels, resulting in superior performance compared to baseline methods. Code and video can be found at https://sites.google.com/view/moddn.

We present MELODI, a novel memory architecture designed to efficiently process long documents using short context windows. The key principle behind MELODI is to represent short-term and long-term memory as a hierarchical compression scheme across both network layers and context windows. Specifically, the short-term memory is achieved through recurrent compression of context windows across multiple layers, ensuring smooth transitions between windows. In contrast, the long-term memory performs further compression within a single middle layer and aggregates information across context windows, effectively consolidating crucial information from the entire history. Compared to a strong baseline - the Memorizing Transformer employing dense attention over a large long-term memory (64K key-value pairs) - our method demonstrates superior performance on various long-context datasets while remarkably reducing the memory footprint by a factor of 8.

Recently, when dealing with high-resolution images, dominant LMMs usually divide them into multiple local images and one global image, which will lead to a large number of visual tokens. In this work, we introduce AVG-LLaVA, an LMM that can adaptively select the appropriate visual granularity based on the input image and instruction. This approach not only reduces the number of visual tokens and speeds up inference, but also improves the overall model performance. Specifically, we introduce the following modules based on LLaVA-NeXT: (a) a visual granularity scaler that includes multiple pooling layers to obtain visual tokens with different granularities; (b) a visual granularity router, which includes a Transformer layer, an MLP layer, and a voter layer, used to select the appropriate visual granularity based on the image and instruction. Furthermore, we propose RGLF, a novel training paradigm that aims at aligning the granularity predicted by the router with the preferences of the LMM, without the need for additional manually annotated data. Extensive experiments and analysis show that AVG-LLaVA achieves superior performance across 11 benchmarks, as well as significantly reduces the number of visual tokens and speeds up inference (e.g., an 85.3% reduction in visual tokens and a 2.53$\times$ increase in inference speed on the AI2D benchmark).

The rapidly growing number and variety of Large Language Models (LLMs) present significant challenges in efficiently selecting the appropriate LLM for a given query, especially considering the trade-offs between performance and computational cost. Current LLM selection methods often struggle to generalize across new LLMs and different tasks because of their limited ability to leverage contextual interactions among tasks, queries, and LLMs, as well as their dependence on a transductive learning framework. To address these shortcomings, we introduce a novel inductive graph framework, named as GraphRouter, which fully utilizes the contextual information among tasks, queries, and LLMs to enhance the LLM selection process. GraphRouter constructs a heterogeneous graph comprising task, query, and LLM nodes, with interactions represented as edges, which efficiently captures the contextual information between the query's requirements and the LLM's capabilities. Through an innovative edge prediction mechanism, GraphRouter is able to predict attributes (the effect and cost of LLM response) of potential edges, allowing for optimized recommendations that adapt to both existing and newly introduced LLMs without requiring retraining. Comprehensive experiments across three distinct effect-cost weight scenarios have shown that GraphRouter substantially surpasses existing routers, delivering a minimum performance improvement of 12.3%. In addition, it achieves enhanced generalization across new LLMs settings and supports diverse tasks with at least a 9.5% boost in effect and a significant reduction in computational demands. This work endeavors to apply a graph-based approach for the contextual and adaptive selection of LLMs, offering insights for real-world applications. Our codes for GraphRouter will soon be released at https://github.com/ulab-uiuc/GraphRouter.