This work investigates the theoretical boundaries of creating publicly-detectable schemes to enable the provenance of watermarked imagery. Metadata-based approaches like C2PA provide unforgeability and public-detectability. ML techniques offer robust retrieval and watermarking. However, no existing scheme combines robustness, unforgeability, and public-detectability. In this work, we formally define such a scheme and establish its existence. Although theoretically possible, we find that at present, it is intractable to build certain components of our scheme without a leap in deep learning capabilities. We analyze these limitations and propose research directions that need to be addressed before we can practically realize robust and publicly-verifiable provenance.

This paper presents SOLOMON, a novel Neuro-inspired Large Language Model (LLM) Reasoning Network architecture that enhances the adaptability of foundation models for domain-specific applications. Through a case study in semiconductor layout design, we demonstrate how SOLOMON enables swift adaptation of general-purpose LLMs to specialized tasks by leveraging Prompt Engineering and In-Context Learning techniques. Our experiments reveal the challenges LLMs face in spatial reasoning and applying domain knowledge to practical problems. Results show that SOLOMON instances significantly outperform their baseline LLM counterparts and achieve performance comparable to state-of-the-art reasoning model, o1-preview. We discuss future research directions for developing more adaptive AI systems that can continually learn, adapt, and evolve in response to new information and changing requirements.

Generative diffusion models in the vision-language domain were the headline topic in the Deep Learning world in 2022. While generating images and videos is definitely a cool playground to try out different models and sampling techniques, we'd argue that In our recent article, we were pondering whether "Denoising Diffusion Is All You Need?". There, we reviewed newest generative models for graph generation (DiGress), molecular conformer generation (EDM, GeoDiff, Torsional Diffusion), molecular docking (DiffDock), molecular linking (DiffLinker), and ligand generation (DiffSBDD). Chroma from Generate Biomedicines allows to impose functional and geometric constraints, and even use natural language queries like "Generate a protein with CHAD domain" thanks to a small GPT-Neo trained on protein captioning; RoseTTaFold Diffusion (RF Diffusion) from the Baker Lab and MIT is packed with the similar functionality also allowing for text prompts like "Generate a protein that binds to X" as well as being capable of functional motif scaffolding, scaffolding enzyme active sites, and de novo protein design. Strong point: 1000 designs generated with RF Diffusion were experimentally synthesized and tested in the lab!

Gibson recently spoke with McKinsey's Lydia The about how AI and machine learning can speed up drug discovery--and what it could mean for patients everywhere. An edited version of their conversation follows. Lydia The: You were part of several biotech firms before you cofounded Generate Biomedicines. Tell us about Generate's potential for revolutionizing medicine. Molly Gibson: Generate is focused on transforming drug discovery in the protein therapeutics space.

One of the more common approaches used to control robots is to programme them with code on detecting objects and feedback loops to specify if they should perform certain tasks. These programmes come at a cost, where reprogramming policies for each task can become time consuming. When provided with natural language instructions, language models at present are highly proficient at not only writing generic code but also generating instructions that let the user control the actions of the robots as well. What if robots could autonomously write their own code to interact with the whole world? One such latest language model, like PaLM, is capable of complex reasoning and has been trained on millions of codes.

DALL-E 2 is one of the most popular AI platforms that offers users the opportunity to create amazing art using text prompts. In this article, we'll show you how to create AI art from scratch as well as edit your own images on the platform. DALL-E 2 is an AI image generation platform that allows users to create images from scratch using text prompts. It runs on an artificial intelligence program called GPT-3, which takes natural language and converts it to images. The platform also allows users to upload their own images and edit them using text prompts to create completely new works of art.

Previous work has shown that there exists a scaling law between the size of Language Models (LMs) and their zero-shot performance on different downstream NLP tasks. In this work, we show that this phenomenon does not hold when evaluating large LMs on tasks with negated prompts, but instead shows an inverse scaling law. We evaluate 9 different tasks with negated prompts on (1) pretrained LMs (OPT & GPT-3) of varying sizes (125M - 175B), (2) LMs further pretrained to generalize to novel prompts (InstructGPT), (3) LMs provided with few-shot examples, and (4) LMs fine-tuned specifically on negated prompts; all LM types perform worse on negated prompts as they scale and show a huge performance gap between the human performance when comparing the average score on both original and negated prompts. By highlighting a critical limitation of existing LMs and methods, we urge the community to develop new approaches of developing LMs that actually follow the given instructions. We provide the code and the datasets to explore negated prompts at this link.

Imagine yourself standing in an art gallery, admiring an abstract print on the wall in front of you. It has bold colors and intentional-looking brush strokes. You wonder what it could mean. Then, you read the label on the wall--and learn the artwork was generated by an algorithm. Maybe you're frustrated that you were "fooled".

Generate Biomedicines is a new kind of therapeutics company – existing at the intersection of machine learning, biological engineering, and medicine – pioneering Generative Biology to create breakthrough medicines where novel therapeutics are computationally generated, instead of being discovered. Generate has built a machine learning-powered biomedicines platform with the potential to generate new drugs across a wide range of biologic modalities. This platform represents a potentially fundamental shift in what is possible in the field of biotherapeutic development. We pursue this audacious vision because we believe in the unique and revolutionary power of generative biology to radically transform the lives of billions, with an outsized opportunity for patients in need. We are seeking collaborative, relentless problem solvers that share our passion for impact to join us!