Welcome to AI-generated encyclopaedia. You can click “Next interesting article” to start using the platform. Contact us if you have any feedback. — Read More
Recent Updates Page 157
Decomposing Language Models Into Understandable Components
Neural networks are trained on data, not programmed to follow rules. With each step of training, millions or billions of parameters are updated to make the model better at tasks, and by the end, the model is capable of a dizzying array of behaviors. We understand the math of the trained network exactly – each neuron in a neural network performs simple arithmetic – but we don’t understand why those mathematical operations result in the behaviors we see. This makes it hard to diagnose failure modes, hard to know how to fix them, and hard to certify that a model is truly safe.
Neuroscientists face a similar problem with understanding the biological basis for human behavior. The neurons firing in a person’s brain must somehow implement their thoughts, feelings, and decision-making. Decades of neuroscience research has revealed a lot about how the brain works, and enabled targeted treatments for diseases such as epilepsy, but much remains mysterious. Luckily for those of us trying to understand artificial neural networks, experiments are much, much easier to run. We can simultaneously record the activation of every neuron in the network, intervene by silencing or stimulating them, and test the network’s response to any possible input.
Unfortunately, it turns out that the individual neurons do not have consistent relationships to network behavior. — Read More
AI Risks
There is no shortage of researchers and industry titans willing to warn us about the potential destructive power of artificial intelligence. Reading the headlines, one would hope that the rapid gains in AI technology have also brought forth a unifying realization of the risks—and the steps we need to take to mitigate them.
The reality, unfortunately, is quite different. Beneath almost all of the testimony, the manifestoes, the blog posts, and the public declarations issued about AI are battles among deeply divided factions. Some are concerned about far-future risks that sound like science fiction. Some are genuinely alarmed by the practical problems that chatbots and deepfake video generators are creating right now. Some are motivated by potential business revenue, others by national security concerns.
The result is a cacophony of coded language, contradictory views, and provocative policy demands that are undermining our ability to grapple with a technology destined to drive the future of politics, our economy, and even our daily lives. — Read More
Google’s RT-2-X Generalist AI Robots: 500 Skills, 150,000 Tasks, 1,000,000+ Workflows
Finding unique features inside LLMs – Interpretability research by Anthropic
The neural networks in large language models show superposition. That means each neuron in the network represents more than one unique feature. Polysemantic neurons compress many rare features of language which is good for performance but makes them harder to understand. You can’t extract those features individually. Anthropic’s new paper tries to extract these hidden features in a human interpretable form. — Read More
Vedeo AI: A Catalog of AI Generated Videos
Political Disinformation and AI
Elections around the world are facing an evolving threat from foreign actors, one that involves artificial intelligence.
Countries trying to influence each other’s elections entered a new era in 2016, when the Russians launched a series of social media disinformation campaigns targeting the US presidential election. Over the next seven years, a number of countries—most prominently China and Iran—used social media to influence foreign elections, both in the US and elsewhere in the world. There’s no reason to expect 2023 and 2024 to be any different.
But there is a new element: generative AI and large language models. These have the ability to quickly and easily produce endless reams of text on any topic in any tone from any perspective. As a security expert, I believe it’s a tool uniquely suited to Internet-era propaganda. — Read More
Adobe teases next-gen AI tools for image editing and object manipulation in a few clicks
Forward-looking: Adobe has gone all-in on AI-based image generation and editing, introducing AI tools across its software stack and perhaps most impressively, doing so in a very timely manner. Ahead of next week’s Adobe Max conference, the company is teasing another AI feature that makes altering photos shockingly easy.
Adobe released a short video teasing Project Stardust, an AI-based photo editing engine that identifies individual objects in photos and allows users to manipulate them as if they were discrete Photoshop layers. The company plans to fully disclose the feature and other AI tools at its Max conference next week. — Read More
Auto-Regressive Next-Token Predictors are Universal Learners
Large language models display remarkable capabilities in logical and mathematical reasoning, allowing them to solve complex tasks. Interestingly, these abilities emerge in networks trained on the simple task of next-token prediction. In this work, we present a theoretical framework for studying auto-regressive next-token predictors. We demonstrate that even simple models such as linear next-token predictors, trained on Chain-of-Thought (CoT) data, can approximate any function efficiently computed by a Turing machine. We introduce a new complexity measure — length complexity — which measures the number of intermediate tokens in a CoT sequence required to approximate some target function, and analyze the interplay between length complexity and other notions of complexity. Finally, we show experimentally that simple next-token predictors, such as linear networks and shallow Multi-Layer Perceptrons (MLPs), display non-trivial performance on text generation and arithmetic tasks. Our results demonstrate that the power of language models can be attributed, to a great extent, to the auto-regressive next-token training scheme, and not necessarily to a particular choice of architecture. — Read More
Rick's Cafe AI 