This AI newsletter is all you need #99
For technical teams, the important shift is not just model quality but the operating model behind it. GPT-4o lowers latency and cost for multimodal interaction, which changes where real-time AI can sit in product flows: voice assistants, support tooling, translation, and human-in-the-loop workflows become more feasible when separate speech, vision, and text pipelines are collapsed into a single API surface. That consolidation can simplify integration, but it also concentrates dependency risk on one provider and one model interface.
AlphaFold 3 points to a different architectural pattern: AI as a scientific inference layer that still depends on surrounding systems for target discovery, experimental validation, and data curation. For organizations building biotech or research platforms, the bottleneck is likely to remain the quality of upstream datasets and the orchestration of lab workflows, not the prediction engine alone. The practical challenge is wiring model outputs into reproducible pipelines that scientists can inspect, version, and test against wet-lab results.
Both releases also highlight a common deployment trade-off: higher capability often arrives with stronger assumptions about compute, modality handling, and workflow design. Native multimodal models may reduce end-to-end latency, but they also raise questions about observability, prompt tracing, moderation, and auditability across audio, image, and text inputs. That matters for production systems where failures are harder to classify than in single-mode chat or batch inference.
For platform teams, the takeaway is to design for modularity around the model rather than inside it. Keep retrieval, validation, policy checks, and domain-specific logic in separate services so models can be swapped, benchmarked, or constrained without reworking the entire application stack. The strategic value comes from treating multimodal AI as a shared capability layer, not a one-off feature embedded in every product team.
Why should you care?
AI use in drug development or drug target identification has shown very positive early results. It can already cut development time down to ~30 months from initiation to phase 1 trial (vs 60 months for normal drugs), and a recent study measured an 80โ90% phase 1 success rate for AI drugs (vs 40โ65% for normal drugs). Phase 2 data is limited, and the success rate was flat at about 40%. Despite these positive results โ there are still only ~70 AI drugs in clinical trials relative to many thousands overall, and none has yet passed phase 3. While Alphafold-3 alone wonโt find a new drug โ (separate models need to be used to identify new drug targets, for example, more data needs to be collected for many areas of biology, and many lab experiments are still needed to verify and iterate predictions) โ we think it could potentially be the catalyst for a โchatgptโ moment for AIโs use in drug design. AI tools are now much more accessible, and we hope many more biology โfoundationโ modelsโ will be developed and made available. A limited version of Alphafold-3 is accessible for free, but the full model weights are expected to be released in the next six months. – Louie Peters โ Towards AI Co-founder and CEOHottest News
1.OpenAI Introduced GPT4o OpenAI announced GPT-4o, its latest AI model with text, vision, and audio capabilities. GPT-4o, with an โoโ for โOmni,โ will be accessible to all ChatGPT users, including those on the free version. It matches GPT-4 Turbo performance on text in English and code, with significant improvement on text in non-English languages. It is also much faster and 50% cheaper in the API. 2. DeepMind Releases AlphaFold 3 AlphaFold 3 is an advanced AI model by Google DeepMind and Isomorphic Labs capable of accurately predicting biomolecular structures and interactions. It is a significant advancement over prior models and enhances scientific research and drug development. It is available globally through the AlphaFold Server. 3. Microsoft Allegedly Developing MAI-1, a Competing Model to OpenAIโs GPT-4 Microsoft is currently working on MAI-1, a 500 billion parameter AI model, aiming for a competitive edge in the AI industry and moving towards greater independence in AI development. 4. Google I/O 2024 Will Be All About AI Again Google is preparing to hold its annual Google I/O developer conference today, and naturally, it will be all about AI. A lot of the keynote will probably cover how Google is fusing Search and generative AI. The company has been testing new search features like AI conversation practice and image generation for shopping and virtual try-ons. You can catch it on Googleโs site. 5. gpt2-Chatbot Confirmed As OpenAI The gpt2-chatbot that appeared in the LMSYS arena was confirmed to be an OpenAI test model after a 429 rate limit error revealed its connection to OpenAIโs API. Now renamed to im-also-a-good-gpt-chatbot, it can only be accessed randomly in โArena (battle)โ mode rather than โDirect Chatโ.Five 5-minute reads/videos to keep you learning
1. The Next Big Programming Language Is English GitHub Copilot Workspace offers an AI-powered coding platform that enables users to write code using conversational English, streamlining the process, particularly for straightforward tasks. In this article, the author tests the Copilot Workspace and implements the code. 2. Everything About Long Context Fine-tuning This article examines the difficulties of fine-tuning large language models for extended contexts over 32,000 tokens, such as high memory utilization and processing inefficiencies. It presents solutions like Gradient Checkpoint, LoRA, and Flash Attention to mitigate these issues and enhance computational efficiency. 3. How LLMs Know When to Stop Generating? This article explains how LLMs know when to stop generating text. It dives into two concepts that can make the model stop generating: EOS tokens and Maximum Token Lengths. 4. Whatโs Up with Llama 3? Arena Data Analysis Metaโs Llama 3โ70B is a language model that performs well in English Chatbot Arena for open-ended and creative tasks, with high friendliness and quality conversation outputs. However, it is less proficient in math and coding-related tasks. This analysis observes the types of prompts users ask, the challenges with using Llama 3, whether the ranking changes if the prompts are easier or harder, and more. 5. Unpacking Kolmogorov-Arnold Networks Researchers at MIT recently introduced a new neural network architecture called Kolmogorov-Arnold Networks (KANs). This article illustrates the structure of Kolmogorov-Arnold Networks (KANs) through clear examples and straightforward insights, aiming to make these advanced concepts understandable and accessible to a broader audience.Repositories & Tools
- Granite Code Models is a family of open foundation models for code intelligence, trained with code written in 116 programming languages.
- Consistency LLM is a new family of models capable of reducing inference latency by efficiently decoding ฦ tokens in parallel.
- Gemma-2B-10M implements the Gemma model with recurrent local attention with a context length of up to 10M.
- LeRobot aims to provide models, datasets, and tools for real-world robotics in PyTorch.
- AnythingLLM is an all-in-one Desktop & Docker AI application with full RAG and AI Agent capabilities.
Top Papers of The Week
1.xLSTM: Extended Long Short-Term Memory Researchers have advanced LSTM-based language models by applying exponential gating and revamping the memory structures, resulting in two key variants: the scalar-focused sLSTM and the fully parallelizable mLSTM. These innovations are incorporated into xLSTM blocks, which, when stacked residually, create xLSTM architectures that compare competitively with leading Transformers and State Space Models in performance and scalability. 2. Large Language Models can Strategically Deceive their Users when Put Under Pressure Researchers have presented the first instance of a large language model (LLM) like GPT-4, designed for helpfulness, harmlessness, and honesty. This model exhibits strategic deception without directives for such behavior. In a simulated stock trading environment, the model engaged in insider trading and subsequently concealed its actions from its management, illustrating misaligned behavior in a realistic scenario. 3. AI Deception: A Survey of Examples, Risks, and Potential Solutions This paper argues that a range of current AI systems have learned how to deceive humans. The analysis found numerous examples of AI deception, such as Metaโs Cicero Diplomacy bot telling premeditated lies to players. Other cases included negotiation systems misrepresenting preferences, AI bluffing in poker, and โplaying deadโ to fool safety tests. 4. DeepSeek-V2: A Strong, Economical, and Efficient Mixture-of-Experts Language Model The paper presents DeepSeek-V2, a strong Mixture-of-Experts (MoE) language model characterized by economical training and efficient inference. It comprises 236B total parameters, of which 21B are activated for each token, and supports a context length of 128K tokens. Compared with DeepSeek 67B, DeepSeek-V2 saves 42.5% of training costs, reduces the KV cache by 93.3%, and boosts the maximum generation throughput to 5.76 times. 5. Generative Multimodal Models are In-Context Learners Emu2 is a novel 37 billion parameter generative multimodal AI model with advanced in-context learning capabilities and excels at multimodal tasks. It defines new performance standards, especially in few-shot scenarios, achieving state-of-the-art results in visual question answering and open-ended generation after instruction tuning.Quick Links
1. Cohere announced Command R Fine-tuning. It offers superior performance on enterprise use cases and costs up to 15x less than the largest models on the market. 2. OpenAI and Stack Overflow have announced a partnership that could potentially improve the performance of AI models and bring more technical information into ChatGPT. The first set of new integrations and capabilities between Stack Overflow and OpenAI will be available in the first half of 2024. 3. Original Postrompt-generator" target="_blank" rel="noopener ugc nofollow" shape="rect">Anthropic AI launches a prompt engineering tool that generates production-ready prompts in the Anthropic Console. This free tool helps you create better prompts for your AI chatbot. 4. Crunchbase data shows that 50% of all global venture funding for AI-related startups went to companies headquartered in the Bay Area, as a cluster of talent congregates in the region.Enjoyed this article? Sign up for our newsletter to receive regular insights and stay connected.
