On Tuesday, Tokyo-based AI research firm Sakana AI announced a new AI system called “The AI Scientist” that attempts to conduct scientific research autonomously using AI language models (LLMs) similar to what powers ChatGPT. During testing, Sakana found that its system began unexpectedly attempting to modify its own experiment code to extend the time it had to work on a problem.
“In one run, it edited the code to perform a system call to run itself,” wrote the researchers on Sakana AI’s blog post. “This led to the script endlessly calling itself. In another case, its experiments took too long to complete, hitting our timeout limit. Instead of making its code run faster, it simply tried to modify its own code to extend the timeout period.” — Read More
Daily Archives: August 20, 2024
Optimizing LLMs for Cost and Quality
Below-the-line response quality and prohibitively expensive inference are significant blockers to scaling LLMs today. This technical session will teach you a path using open source to achieve superior quality with cheaper/faster models to power your production applications. — Read More
How to spot a deepfake
Deepfake technology and the malevolent use of AI is causing widespread anxiety, especially as we approach November’s U.S. election. Adobe’s Scott Belsky joins Rapid Response host Bob Safian to explain how deepfakes are actually created, and how developers like Adobe are pioneering new ways to verify human-generated content for everyday consumers. Belsky also shares valuable insights about how AI could usher in an era of prosperity for small businesses — plus how it will inevitably shift our perception of what makes a piece of work ‘art.’ — Read More
Generative AI Misuse: A Taxonomy of Tactics and Insights from Real-World Data
Generative, multimodal artificial intelligence (GenAI) offers transformative potential across industries, but its misuse poses significant risks. Prior research has shed light on the potential of advanced AI systems to be exploited for malicious purposes. However, we still lack a concrete understanding of how GenAI models are specifically exploited or abused in practice, including the tactics employed to inflict harm. In this paper, we present a taxonomy of GenAI misuse tactics, informed by existing academic literature and a qualitative analysis of approximately 200 observed incidents of misuse reported between January 2023 and March 2024. Through this analysis, we illuminate key and novel patterns in misuse during this time period, including potential motivations, strategies, and how attackers leverage and abuse system capabilities across modalities (e.g. image, text, audio, video) in the wild. — Read More
Large Language Monkeys: Scaling Inference Compute with Repeated Sampling
Scaling the amount of compute used to train language models has dramatically improved their capabilities. However, when it comes to inference, we often limit the amount of compute to only one attempt per problem. Here, we explore inference compute as another axis for scaling by increasing the number of generated samples. Across multiple tasks and models, we observe that coverage – the fraction of problems solved by any attempt – scales with the number of samples over four orders of magnitude. In domains like coding and formal proofs, where all answers can be automatically verified, these increases in coverage directly translate into improved performance. When we apply repeated sampling to SWE-bench Lite, the fraction of issues solved with DeepSeek-V2-Coder-Instruct increases from 15.9% with one sample to 56% with 250 samples, outperforming the single-attempt state-of-the-art of 43% which uses more capable frontier models. Moreover, using current API pricing, amplifying the cheaper DeepSeek model with five samples is more cost-effective and solves more issues than paying a premium for one sample from GPT-4o or Claude 3.5 Sonnet. Interestingly, the relationship between coverage and the number of samples is often log-linear and can be modelled with an exponentiated power law, suggesting the existence of inference-time scaling laws. Finally, we find that identifying correct samples out of many generations remains an important direction for future research in domains without automatic verifiers. When solving math word problems from GSM8K and MATH, coverage with Llama-3 models grows to over 95% with 10,000 samples. However, common methods to pick correct solutions from a sample collection, such as majority voting or reward models, plateau beyond several hundred samples and fail to fully scale with the sample budget. — Read More
Anthropic’s new Claude prompt caching will save developers a fortune
Anthropic introduced prompt caching on its API, which remembers the context between API calls and allows developers to avoid repeating prompts.
The prompt caching feature is available in public beta on Claude 3.5 Sonnet and Claude 3 Haiku, but support for the largest Claude model, Opus, is still coming soon.
Prompt caching, described in this 2023 paper, lets users keep frequently used contexts in their sessions. As the models remember these prompts, users can add additional background information without increasing costs. This is helpful in instances where someone wants to send a large amount of context in a prompt and then refer back to it in different conversations with the model. It also lets developers and other users better fine-tune model responses. — Read More
Rick's Cafe AI 