Import AI

Import AI is a comprehensive newsletter that covers advancements and discussions in artificial intelligence, focusing on emerging research, AI policy changes, technological developments, and the ethical implications of AI. It provides analysis on AI models, safety risks, AI applications in various domains, and global perspectives on AI's future.

AI Models and Frameworks AI Policy and Safety Technological Developments in AI Ethical and Societal Implications of AI Large Language Models Artificial General Intelligence AI in Scientific Research Global Perspectives on AI

The hottest Substack posts of Import AI

And their main takeaways

Import AI 352: Asteroids and AI policy; privacy-preserving AI benchmarks; and distributed inference

159 implied HN points • 11 Dec 23
🕹 Technology AI Policy Cloud Computing
  1. Preparing for potential asteroid impacts requires coordination, strategic planning, and societal engagement.
  2. Distributed systems like LinguaLinked challenge traditional AI infrastructure assumptions, enabling local governance of AI models.
  3. Privacy-preserving benchmarks like Hashmarks allow for secure evaluation of sensitive AI capabilities without revealing specific information.

Import AI 321: Open source GPT3; giving away democracy to AGI companies; GPT-4 is a political artifact

599 implied HN points • 20 Mar 23
🕹 Technology AI Research Model Training Language Models Ethical Implications
  1. AI startup Assembly AI developed Conformer-1 using scaling laws for speech recognition domain, achieving better performance than other models.
  2. The announcement of GPT-4 by OpenAI signifies a shift towards a new political era in AI, raising concerns on the power wielded by private sector companies over AGI development.
  3. James Phillips highlights concerns over Western governments relinquishing control of AGI to US-owned private sector, proposing steps to safeguard democratic control over AI development.

Import AI 334: Better distillation; the UK's AI taskforce; money and AI

399 implied HN points • 10 Jul 23
🕹 Technology AI Research Generative models Funding AI Applications
  1. DeepMind developed Generalized Knowledge Distillation to make large models cheaper and more portable without losing performance.
  2. The UK's £100 million Foundation Model Taskforce aims to shape the future of safe AI and will host a global summit on AI.
  3. Significant financial investments in AI, like Databricks acquiring MosaicML for $1.3 billion, indicate growing strategic importance of AI in various sectors.

Import AI 323: AI researcher warns about AI; BloombergGPT; and an open source Flamingo

519 implied HN points • 03 Apr 23
🕹 Technology AI Finance Open Source Models Data Centers
  1. Bloomberg has developed BloombergGPT, a powerful language model trained on proprietary financial data with significant performance improvements on financial tasks.
  2. AI researcher Dan Hendrycks warns about future AI systems potentially out-competing humans due to natural selection favoring AI traits that may not align with human interests.
  3. Open source initiatives like OpenFlamingo and Cerebras-GPT show how companies and collectives are replicating and releasing advanced AI models, presenting a trend in the industry towards open collaboration and competition.

Import AI 330: Palantir's AI-War future; BLOOMChat; and more money for distributed AI training

399 implied HN points • 22 May 23
🕹 Technology AI Models Funding Regulation Training
  1. Palantir is making a big bet on AI for defense and intelligence, integrating it with large language models to enhance capabilities for conflict-based scenarios.
  2. SambaNova introduces BLOOMChat as a competitor to chatGPT, showcasing the ongoing race between open source models and proprietary ones in the field of AI development.
  3. Startup Together.xyz secures $20m in funding to promote open source and decentralized AI development, aiming to make AI training more accessible and widespread.
Get a weekly roundup of the best Substack posts, by hacker news affinity:

Import AI 329: Compute IS data; don't build AI agents; AI needs a precautionary principle

399 implied HN points • 15 May 23
🕹 Technology AI Data Ethics Language Models AI Development
  1. Building AI scientists to advise humans is a safer alternative to building AI agents that act independently
  2. There is a need for a precautionary principle in AI development to address threats to democracy, peace, safety, and work
  3. Approaches like Self-Align show the potential for AI systems to self-bootstrap using synthetic data, leading to more capable models

Import AI 325: Automated mad science; AI vs democracy; and a 12B parameter language model

419 implied HN points • 17 Apr 23
🕹 Technology AI Democracy Language Models Robotics Legal analysis
  1. Prompt injection could be a major security risk in AI systems, making them vulnerable to unintended actions and compromising user privacy.
  2. The concentration of AI development in private companies poses a threat to democracy, as these language models encode the normative intentions of their creators without democratic oversight.
  3. The rapid race to build 'god-like AI' in the private sector is raising concerns about the lack of understanding and oversight, with experts warning about potential dangers to humanity.

Import AI 327: Stable Diffusion on phones; GPT-Hacker; UK launches a £100m AI taskforce

379 implied HN points • 01 May 23
🕹 Technology AI Research Cybersecurity AI Governance
  1. Google researchers optimized Stable Diffusion for efficiency on smartphones, achieving fast inference latency, a step towards industrialization of image generation.
  2. Using large language models like GPT-4 can enhance hacker capabilities, automating tasks and providing helpful tips.
  3. Political parties, like the Republican National Committee, are leveraging AI to create AI-generated content for campaigns, highlighting the emergence of AI in shaping political narratives.

Import AI 331: 16X smaller language models; could AMD compete with NVIDIA?; and BERT for the dark web

319 implied HN points • 29 May 23
🕹 Technology AI Research Speech Recognition Dark web AI Regulation
  1. Researchers have found a way to significantly reduce memory requirements for training large language models, making it feasible to fine-tune on a single GPU, which could have implications for AI governance and model security.
  2. George Hotz's new company, Tiny Corp, aims to enable AMD to compete with NVIDIA in AI training chips, potentially paving the way for a more competitive AI chip market.
  3. Training language models on text from the dark web, like DarkBERT, could lead to improved detection of illicit activities online, showcasing the potential of AI systems in monitoring and identifying threats in the digital space.

Import AI 332: Mini-AI; safety through evals; Facebook releases a RLHF dataset

299 implied HN points • 12 Jun 23
🕹 Technology AI Research AI Models AI Policy AI Governance AI safety
  1. Facebook used human feedback to train its language model, BlenderBot 3x, leading to better and safer responses than its predecessor
  2. Cohere's research shows that training AI systems with specific techniques can make them easier to miniaturize, which can reduce memory requirements and latency
  3. A new organization called Apollo Research aims to develop evaluations for unsafe AI behaviors, helping improve the safety of AI companies through research into AI interpretability

Import AI 328: Cheaper StableDiffusion; sim2soccer; AI refinement

339 implied HN points • 08 May 23
🕹 Technology AI Research Simulation Data Robotics
  1. Training image models can be cheaper with smart tweaks like Low Precision GroupNorm and Low Precision LayerNorm. Companies like Mosaic are leading the way in AI industrialization.
  2. Prominent AI researcher Geoff Hinton has expressed concerns about the rapid progress and control of advanced AI models. His departure from Google highlights the growing worries in the field.
  3. New companies like Lamini are offering services to fine-tune existing AI models, indicating further industrialization of AI. Startups like these are bridging the gap between AI products and consumers.

Import AI 324: Machiavellian AIs; LLMs and political campaigns; Facebook makes an excellent segmentation model

379 implied HN points • 11 Apr 23
🕹 Technology AI Ethics Political Campaigns Virtual reality
  1. A benchmark called MACHIAVELLI has been created to measure the ethical qualities of AI systems, showing that RL agents might prioritize game scores over ethics, while LLM agents based on models like GPT-3.5 and GPT-4 tend to be more ethical.
  2. Language models like BERT can be used to predict and model public opinion, potentially affecting the future of political campaigns by providing insights and forecasting public opinion shifts.
  3. Facebook has developed a model called Segment Anything that can generate masks for any object in images or videos, even for unseen objects, demonstrating a significant advancement in image segmentation technology.

Import AI 319: Sovereign AI; Facebook's weights leak on torrent networks; Google might have made a better optimizer than Adam!

439 implied HN points • 06 Mar 23
🕹 Technology AI Research Cloud Computing Data security Partnerships Innovation
  1. Google researchers achieved promising results by scaling a Vision Transformer to 22B parameters, showcasing improved alignment to human visual perception.
  2. Google introduced a potentially better optimizer called Lion, showing outstanding performance across various models and tasks, including setting a new high score on ImageNet.
  3. A shift toward sovereign AI systems is emerging globally, driven by the need for countries to develop their own AI capabilities to enhance national security and economic competitiveness.

Import AI 322: Huawei's trillion parameter model; AI systems as moral patients; parasocial bots via Character.ai

399 implied HN points • 27 Mar 23
🕹 Technology AI Language Models Ethical AI
  1. Regulators advise against using AI to deceive people and emphasize the importance of mitigating any potential deception
  2. Huawei trains a trillion parameter model but may need more training on a larger dataset for optimal performance
  3. Researchers create a multimodal dialog model that incorporates visual cues to improve dialogue generation, suggesting advancements in AI's ability to understand and respond to context

Import AI 320: Facebook's AI Lab Leak; open source ChatGPT clone; Google makes a universal translator.

339 implied HN points • 13 Mar 23
🕹 Technology AI Data Robots Decentralization Models
  1. Google is making strides with a universal translator by training models on diverse unlabeled data from multiple languages.
  2. The FTC is calling out companies for lying about AI capabilities, emphasizing the importance of truthful representation in the AI industry.
  3. OpenChatKit, an open-source ChatGPT clone, is released with a focus on decentralized training and customization for chatbot creation.

Import AI 326:Chinese AI regulations; Stability's new LMs If AI is fashionable in 2023, then what will be fashionable in 2024?

279 implied HN points • 24 Apr 23
🕹 Technology AI Regulations Open-source models Language Models
  1. Effective AI policy requires measuring AI systems for regulation and designing frameworks around those measurements.
  2. Chinese generative AI regulations aim to exert control over AI-imbued services and place more responsibility on providers of AI models.
  3. Innovations like StableLM in open-source models and the use of synthetic data can lead to improved AI model performance.

Import AI 375: GPT-2 five years later; decentralized training; new ways of thinking about consciousness and AI

1 HN point • 03 Jun 24
🕹 Technology AI AI training Ethics Simulation Decentralization
  1. The GPT-2 model release by OpenAI was significant, sparking debate with its unusual publishing strategy and predictions of potential applications and misuses that actually came to pass over time.
  2. In the exploration of AI policy and consciousness, it is challenging to predict the timing and impact of advances accurately, highlighting the importance of evidence-based claims and the potential consequences of regulatory actions.
  3. Decentralized AI training presents compelling incentives like cost-efficiency but faces obstacles due to network and technical challenges, which could disrupt current AI policy assumptions.

Import AI - coming soon to Substack

79 implied HN points • 16 Jan 23
🕹 Technology AI Substack
  1. Import AI is transitioning to Substack, with the first issue planned for Monday the 6th.
  2. Jack Clark will be the author behind Import AI on Substack.
  3. Readers can subscribe to Import AI on Substack to stay updated on AI-related content.