The week in AI: Trump allies draft an AI order to launch ‘Manhattan Projects’ for defense, repeal Biden's AI Executive Order

In partnership with

Welcome to The Dispatch! We are the newsletter that keeps you informed about AI. Each Thursday, we aggregate the major developments in artificial intelligence - we pass along the news, useful resources, tools and services; we highlight the top research in the field as well as exciting developments in open source. Even if you aren’t a machine learning engineer, we’ll keep you in touch with the most important developments in AI.

NEWS & OPINION

Trump allies draft AI order to repeal Biden’s Executive Order and launch ‘Manhattan Projects’ for defense

-------------------------

Former President Donald Trump's allies are drafting an AI executive order that could significantly reshape the US approach to artificial intelligence development and regulation. The proposed framework, titled "Make America First in AI, " represents a stark departure from the Biden administration's more cautious stance and favors rapid innovation with reduced oversight. Here are the key elements from the draft, exclusively shared with the Washington Post:

• Military AI development: Proposes ‘Manhattan Projects’ for AI to advance military capabilities. The terminology here should say it all; the US is already heavily invested in AI for military purposes, and a hyper-focused military lens on the technology could escalate global AI arms races and deepen ethical concerns about AI in warfare.

• Regulatory rollback: Calls for immediate review and potential elimination of existing AI regulations, including repealing Biden's AI Executive Order. Per the 2024 GOP platform statement:
  
  We will repeal Joe Biden's dangerous Executive Order that hinders AI Innovation, and imposes Radical Leftwing ideas on the development of this technology. In its place, Republicans support AI Development rooted in Free Speech and Human Flourishing.

• Industry-led oversight: Suggests creating industry-led agencies to evaluate AI models and safeguard systems from foreign threats. This approach would likely benefit tech companies already collaborating with the Pentagon on AI projects, such as Palantir, Anduril, and Scale AI. Executives from all of these firms have reportedly expressed support for Trump.

• "Make America First in AI" initiative: Emphasizes maintaining US leadership in AI, particularly in competition with China. The potential effectiveness of this type of nationalistic approach in the current AI landscape is debatable - but it centers around trade restrictions on advanced AI chips and limiting China’s access and funding towards the technology. There’s a huge black market for advanced chips in China (and the equipment required to make them), and the country already leads the world in generative AI adoption. Further stifling China’s AI development would be a major challenge.

Additionally, Trump’s announcement of running mate J.D. Vance has reportedly energized supporters of rapid AI development. Vance, known for his venture capital background and ties to Silicon Valley figures, has voiced opposition to AI regulation but has also been critical of the power Big Tech companies currently hold.

Microsoft releases first details of SpreadsheetLLM - a new model that has the "potential to transform spreadsheet data management and analysis”

-------------------------

Last Friday, Microsoft published its research on SpreadsheetLLM - and it looks to be a significant step up in applying artificial intelligence to data analysis and management. State-of-the-art LLMs like ChatGPT, Claude, and Gemini have some major limitations in their interactions with spreadsheets. Spreadsheets have a complex two-dimensional structure, flexible layouts, and varied formatting - factors that contribute to poor task comprehension, unreliable outputs, and excessive token use. SpreadsheetLLM takes on these challenges with a ‘SheetCompressor’ framework, which employs structural anchors, inverse index translation, and data-format-aware aggregation. This enhances the model's ability to understand and interact with spreadsheets; and as the first model of its kind, it paves the way for more sophisticated AI-driven data analysis.

The linked article starts off with a line about accountants and data analysts fearing for their future job prospects - it’s more likely (for now, at least!) that models like SpreadsheetLLM will simply augment human capabilities, hopefully allowing professionals to focus on higher-level data analysis and management. SpreadsheetLLM offers practical applications, and can already:

• Simplify data cleaning, formatting, and aggregation

• Automate routine data analysis tasks

• Provide intelligent insights and recommendations

MORE IN AI THIS WEEK

• AI has become a technology of faith

• Universities don’t want AI research to leave them behind

• TSMC says AI chip constraints to last through 2025

• The sperm whale 'phonetic alphabet' revealed by AI

• ‘Visual’ AI models might not see anything at all

• Apple, Nvidia, and Anthropic used thousands of swiped YouTube videos to train their AI

• Apple at all-time high after Morgan Stanley calls stock 'top pick' for AI efforts

• OpenAI illegally barred staff from airing safety risks, whistleblowers say

• OpenAI’s 5-level roadmap to AGI - they say they’re on step 2

• OpenAI's new 'Project Strawberry' could give ChatGPT more freedom to search the web and solve complex problems

• AMD to acquire Finnish startup Silo AI for $665 mln to step up in AI race

• Scale is partnering with Amazon/AWS to boost enterprise and government adoption of generative AI

• Elon Musk confirms Tesla’s robotaxi unveiling event delayed due to design change

• Japan's SoftBank acquires British AI chipmaker Graphcore

• Rep. Jennifer Wexton (D-Va.), confronting degenerative disease, finds her voice through AI

FREE AI & ChatGPT Masterclass to automate 50% of your workflow

More than 300 Million people use AI across the globe, but just the top 1% know the right ones for the right use-cases.

Join this free masterclass on AI tools that will teach you the 25 most useful AI tools on the internet – that too for $0 (they have 100 free seats only!)

Get it now for absolutely free! (for first 100 users only) 🎁

This masterclass will teach you how to:

• Build business strategies & solve problems like a pro

• Write content for emails, socials & more in minutes

• Build AI assistants & custom bots in minutes

• Research 10x faster, do more in less time & make your life easier

You’ll wish you knew about this FREE AI masterclass sooner 😉

Register & save your seat now! (valid for next 24 hours only!)

TECHNICAL, RESEARCH & OPEN SOURCE

• You can now run your own AI cluster at home using everyday devices

• Can ChatGPT do data science?

• Generative AI enhances individual creativity but reduces the collective diversity of novel content

• Mistral open-sources Codestral Mamba (architecture)

• (Microsoft research)‘Arena Learning’: a new AI-powered method for post-training LLMs using simulated chatbot battles that significantly increases performance, efficiency

• Claude Dev: an autonomous software engineer right in your IDE

• Anthropic introduces fine-tuning for Claude 3 Haiku available in Amazon Bedrock

• Meta to drop Llama 3 400b next week - here’s why you should care

That’s all for this week! We’ll see you next Thursday.