A big week for AI policy
Plus: The Boom of 3D Content Generation
Happy Friday! “AI” was officially named word of the year by Collins Dictionary—its use in the news and on social media quadrupled in 2023, according to Collins. Other words that made the list this year? “Nepo baby” and “debanking.”
Speaking of AI and news, here are this week’s biggest AI headlines.
AI Policy is Heating Up
AI safety was the topic of the week on both sides of the pond. Here’s what you need to know:
President Biden’s executive order (EO): With a new EO issued on Tuesday, President Biden seeks to establish “new standards” for AI safety and security. Within its 100 pages, this EO sets out that…
Developers of “the most powerful AI systems” (think: Google, OpenAI, etc.) must share the results of their safety tests and risk assessments with the government before public release.
Equity and civil rights are protected by preventing AI from exacerbating discrimination in areas like criminal justice, healthcare, and housing.
…and much more, from supporting AI workers to protecting against the risk of AI-engineered biological materials.
There were many positive reactions that applauded the EO for addressing government oversight, while leaving room for technological innovation. Still, others argue that the language is too vague. The EO is the biggest action a major leader has taken on AI to date—but it still requires Congress to implement legislation for any enforcement.
AI Safety Summit: The U.K. hosted the first major global summit on AI safety on Wednesday and Thursday. The agenda? Understand the threats that AI poses and discuss how they can be mitigated by international collaboration.
The Summit’s biggest achievement was the signing of the Bletchley Declaration. It’s the first international declaration on AI, with the 28 attending countries agreeing that AI poses a potentially catastrophic risk to humanity.
Zoom out: If this week showed us anything, it’s that regulators worldwide are getting serious about managing the risks and leveraging the potential of advanced AI models. While President Biden’s EO spearheads the conversation around national AI policy, the UK’s Summit stresses the need for a global approach to managing AI risk.
Today’s Future Tools brought to you by:
The AI Infrastructure Alliance’s Guide to Agents, LLMs, and Smart Apps
Anyone can create the next big AI app.
We’re at a critical point in AI development, and if you’re thinking of building—now’s the time.
In this article, we break down exactly where we are in the evolution of AI. Plus, we share the all-in-one guide you need to navigate everything from prompt engineering to major frameworks, to vector DBs, fine-tuning, open source and closed source generative models, legal implications, common app design patterns, LLM logic and reasoning and more.
Last year, ChatGPT created the current generative AI explosion, but it’s apps that do real work for people that will unleash the coming wave of next-gen AI software, and this guide helps get you ready to write and use the apps of tomorrow.
🗞️ Other AI News & Articles
The Boom of 3D Content Generation
After a flurry of innovation in image generation, AI developers are ready to tackle the next big thing: 3D content generation. While AI-powered 3D content generation has been around for a hot minute (since 3DFY’s release in 2017), realistic and accurate 3D content is still relatively new—and AI developers are racing to innovate.
This week, we saw lots of exciting innovation in 3D content generation:
Stable 3D: Stability AI just launched its first 3D content generation app: Stable 3D. Designed to enable non-experts to generate “draft-quality” 3D models, Stable 3D builds on top of the Stable Diffusion model to output 3D content in the “.obj” file format—enabling easy edits via most standard 3D modeling tools. Using a view-dependent diffusion model, Stable 3D can generate 3D images that are accurate and realistic from any angle.
DreamCraft3D: Researchers from Tsinghua University presented DreamCraft3D—a model that produces 3D content from a 2D reference image or text description. It’s a hierarchical model, which means that it works by first generating a rough 2D version of the image, and then gradually refining it until it produces a highly accurate and realistic 3D model.
DreamCraft3D aims to solve the common problem of geometric inconsistency in generated 3D objects—and achieves this through a breakthrough process called “score distillation sampling.” If you’re interested in the nitty gritty of this process, check out this paper.
A new text-to-3D model: Deemos, a Chinese software development studio, officially introduced Rodin Gen-1—a new generative AI model that generates complex 3D shapes and highly realistic textures (specifically PBR textures, which mimic light scattering and reflection) from text descriptions.
Text-to-3D via Discord: Luma just released a beta research preview of their newest text-to-3D model, Genie, which allows you to create 3D models within Discord.
The takeaway: Historically, 3D content generation apps have been challenging to produce due to a lack of 3D training data and inefficient algorithms. But after the successful launch of many 2D image generators, the foundation has been built for 3D development…and I’m predicting that this is just the beginning of an even bigger wave.
Apple Finally Steps into the AI Ring
While every other company in tech is slapping “AI” on every imaginable product, Apple hasn’t mentioned the term once in any public announcement or statement…like, ever.
But that just changed.
The details: Apple unveiled its new M3 chip series at its hardware event this week, mentioning “AI” twice in the official presentation. The M3 chips have an increased memory capacity to Apple’s other chip series, allowing developers to run large LLMs with billions of parameters from a 14-inch laptop.
This is BIG news: The unveiling of M3 chips in tandem with Apple’s new MacBook Pros wasn’t your normal product announcement: By mentioning AI for the first time, the company is explicitly positioning the new MacBooks as a tool for AI developers. This is Apple’s official entrance into the AI race, heating up the chip rivalry with Big Tech giants like Google and Amazon.
Google agreed to invest up to $2 billion in Anthropic—just weeks after Amazon’s $4 billion commitment.
Midjourney, Stability AI, and Deviant Art scored a court victory in an AI copyright case by artists.
Microsoft started rolling out Windows 11 with integrated Copilot.
Instagram is reportedly developing a customizable “AI friend”.
Quora’s AI platform Poe introduced an AI chatbot creator economy.
Scarlett Johansson hits AI app with legal action for cloning her voice in an ad.
📺️ This Week’s Videos to Watch:
More important AI news: Dive deeper into this week’s hottest AI news stories (because yes, there are even more) in my latest YouTube video:
Tool hack: Yes, the Swiss army knife of AI tools exists. Watch me play around with it here:
Midjourney: ICYMI, Midjourney has a new website, and people are loving it. Here’s a walkthrough:
And there you have it! I’ll be back in your inbox on Wednesday with five new AI tools. In the meantime, Happy Halloween to all the avid trick-or-treaters out there! Stay safe and keep an eye open for any AI-themed costumes. :)
If you haven’t already, be sure to follow me over on X. I like to share news and AI updates as they happen over there. And don't forget to check out all of the newest tools we've just added on Future Tools!
You rock! See ya next week. :)
P.S. This newsletter is 100% written by a human. Okay, maybe 96%.