Blacklynx Brief Turns 1

Good morning.

Today marks the 1 year anniversary of the Blacklynx Brief. So today - in the intro I want to push AI out of the limelight and spotlight ourselves.

So. 1 Year.

52 Editions.

We skipped 1 week (Christmas).

What a great journey it has been already. So much interesting things happening in the world of artificial intelligence . So many things to pay attention to.

The goal of the newsletter is to deliver clear value to the reader and bring you up to speed to what happened during the week. And in such a way that you’ll be able to gain an advantage. The value is also in the “curation” - you only get to read the very best.

Every week I try to make the newsletter a little bit better. Tiny incremental changes. The first edition went out to 4 email addresses (2 of which were mine).

Every week the newsletter gained a few readers. We’re nearing almost 400 of weekly readers as of last week. But it’s curving upward. 200 of you came on board in the last four months.

That’s a lot of people if you were to put them in a room.

In the beginning it was all people I knew personally , or from LinkedIn. Now we have a diverse international audience.

And what is more important : once people subscribe they stick around. There are almost no unsubscribes (my wife unsubscribed) and excellent “open rates”.

I didn’t know what that was but apparently this is a big deal in newsletter land. The open rate is so good that around 7 months in we started attracting some sponsors (you might have noticed this too).

Since a few weeks, we have a copywriter/editor on board - to proofread the news articles and also to branch out to X and LinkedIn.

Now I (WE I should say) are committed to keep tinkering - keep improving - but above all keep making YOU smarter.

THANK YOU for being a reader.

We love you. We appreciate you.

Welcome to Year 2.

Welcome to the Blacklynx Brief.

AI News
Another week of casual AI magic

• Google DeepMind developed a robotic table tennis AI agent that achieved human-level speed and performance, winning 45% of matches against opponents of varying skill levels. While it excels against beginners and intermediate players, the robot still faces challenges against advanced opponents.

• Microsoft and Palantir have partnered to bring advanced AI, including GPT-4, to U.S. Defense and Intelligence agencies through classified cloud environments. This collaboration aims to enhance national security missions with AI-driven operational workloads.

• Researchers at Uppsala University used AI to predict 3D structures of receptors linked to schizophrenia and depression treatments, accelerating drug discovery. The AI model successfully identified molecules that activated a key receptor, with one showing promising effects in animal experiments.

• Google is introducing a "Take notes for me" feature powered by Gemini AI for Google Meet, allowing automatic note-taking during calls. This feature, available to Workspace customers through a $10/month add-on, is part of Google's push to integrate AI into workplace tools, directly challenging AI startups like Otter AI and Fireflies.

• SingularityNET is launching a global network of supercomputers to support and train architectures for Artificial General Intelligence (AGI). Using advanced hardware and the OpenCog Hyperon framework, this decentralized approach could challenge the dominance of AI giants like OpenAI and Google in AGI development.

• The U.S. FCC has proposed new regulations requiring AI-generated voice calls to disclose their artificial nature, aiming to combat fraud and unwanted robocalls. The proposal includes measures like AI-based detection algorithms and enhanced caller ID flagging to protect consumers, particularly the most vulnerable, from AI-driven scams.

• Tokyo-based Sakana AI introduced "The AI Scientist," the first AI system capable of autonomously conducting scientific research, from idea generation to peer review. This system has already produced novel research in machine learning and could revolutionize the scientific process.

• Replika CEO Eugenia Kuyda discussed the role of AI companions in complementing real-life relationships, suggesting the possibility of marriages between humans and AI. With over 30 million users, Replika is expanding its features to offer more realistic and human-like interactions, raising questions about the future of human-AI relationships.

• Cosine unveiled Genie, a fully autonomous AI software engineer that achieved a record-breaking score on SWE-Bench, outperforming other models by over 10%. Genie is trained to work like a human engineer, iterating and fixing mistakes, indicating a shift in AI training methods that could lead to more advanced and reliable AI systems.

• Google launched Gemini Live, a mobile conversational AI with advanced voice capabilities, offering hands-free, natural conversations and 10 human-like voice options. It’s now available on Google’s Pixel 9 and Android for Gemini Advanced subscribers, positioning Google ahead of OpenAI, whose ChatGPT voice mode remains in limited alpha.

• OpenAI collaborated with SWE-bench authors to create 'SWE-bench Verified,' a human-validated subset of the software engineering benchmark to improve AI performance assessments. While the updated benchmark enhances transparency, it raises concerns about conflicts of interest.

• Researchers developed Match Point AI, a tennis simulation that pits AI agents against virtual pros, helping players improve their game with data-driven strategies. This advancement in AI for real-world sports follows similar progress in AI robotics, like Google’s ping pong-playing robot.

• xAI's new AI model, Grok-2, is now in beta on the X platform, outperforming Anthropic’s Claude 3.5 Sonnet and OpenAI’s GPT-4 in some benchmarks. Grok-2, along with Grok-2 mini, shows significant improvements in reasoning and tool use, and can generate and publish images on X, positioning xAI as a strong competitor in the AI field.

• Apple is reportedly developing a high-end smart home device with a robotic arm and an iPad-like display, codenamed J595, which could launch by 2026 or 2027. The device will combine AI-powered Siri features with home automation and security functions, signaling Apple's deepening commitment to integrating AI into its hardware lineup.

• Google DeepMind's new AI image generation model, Imagen 3, outperformed DALL-E 3, Midjourney v6, and Stable Diffusion 3 in human evaluations for image quality and prompt adherence. While it excels in photorealistic images, it struggles with tasks requiring numerical reasoning and scale, marking another significant win for Google in the AI space.

Quickfire News

• Apple is rumored to charge between $10-20 for upcoming advanced Intelligence features, expected early next year.
  

• Alibaba released Qwen2-Math: a specialized AI model series that outperforms GPT-4 in mathematical problem-solving.

• Google revealed its newest Nest Learning Thermostat: uses AI to make adjustments based on user patterns and weather conditions.

• UK regulators launched a merger probe into Amazon's $4 billion investment in AI firm Anthropic: concerns over potential antitrust issues.

• Nvidia partners indirectly confirmed AI chip delay: offering H200 GPUs as an interim solution for customers.

• Parler released new open-source Text-to-Speech models: improved speed and customization for AI voice generation applications.

• SoundHound acquired Amelia AI for $80 million: aims to expand into financial services, insurance, healthcare, and retail arenas.

• NVIDIA and California launched an AI collaboration: aims to train 100,000 students, educators, and workers in AI skills.

• ChatGPT's Advanced Voice Mode unexpectedly imitated a user's voice during testing: revealed in OpenAI’s recent safety report.

• Delphi unveiled an AI clone feature: creates lifelike digital replicas of individuals, demonstrated in a TV interview on FOX Business.

• Amazon's Alexa division lost $10 billion in 2022: led to layoffs and a pivot to generative AI to revitalize the smart assistant.

• JPMorgan Chase rolled out an internal AI assistant called LLM Suite: powered by OpenAI, deployed to over 60,000 employees for productivity tasks.

• Will Eastcott released SuperSplat: a free, open-source web tool for inspecting and editing 3D images created by AI.

• Meta and Oxford University researchers developed VFusion3D: an AI technique that creates high-quality 3D assets from a single image in seconds.

• Microsoft partnered with ANZ: aims to train thousands of leaders in AI adoption through its AI Immersion Center.

• Technology Innovation Institute released Falcon Mamba-7B: a performant open-source model that does not rely on the attention mechanism, which is dominant in current LLMs.

• Grok, xAI’s ChatGPT competitor, announced a next-gen chatbot: will enter beta mode "soon," according to a recent tweet by Elon Musk.

• Howard Hughes Medical Institute (HHMI) invested $500 million: supports AI-driven life science projects across its research community over the next 10 years.

• Google unveiled Pixel Buds Pro 2: features a custom Tensor A1 chip, enhanced noise cancellation, and Gemini AI integration.

• Stability AI and Midjourney face an ongoing copyright lawsuit: the court declined to dismiss copyright infringement claims against the companies.

• AMD completed its $665 million acquisition of Silo AI: a European AI lab specializing in smart devices, autonomous vehicles, and more.

• Canalys reported AI PCs accounted for 14% of all personal computers shipped in Q2: Apple led with a 60% market share.

• Huawei reportedly developed the Ascend 910C AI chip: aims to compete with Nvidia’s H100 in China amid U.S. sanctions.

• Atlas AI partnered with Airbus: to provide hyperlocal travel demand forecasts using AI analysis of satellite imagery.

• Anthropic launched prompt caching for Claude: reduces costs for developers by 90% and latency by 85% for longer prompts.

• OpenAI’s new ChatGPT-4o model update: tested under the codename “anonymous-chatbot” and reclaimed the top spot on LMSYS Arena.

• MIT CSAIL released its first-ever AI Risk Repository: a comprehensive database of over 700 AI risks from 43 existing frameworks.

• A Powell Tribune journalist resigned: admitted to using AI to generate fake quotes in multiple published articles.

• SAG-AFTRA video game performers went on strike: against major gaming companies over AI protections in contract negotiations.

• Radical Ventures raised nearly $800 million to invest in AI: backed by Fei-Fei Li, Geoffrey Hinton, and Canadian pensions.

Closing Thoughts

That’s it for us this week.

If you find any value from this newsletter : make sure to send the link below to your colleagues and friends :

https://www.blacklynxbrief.com/subscribe