. . AI: Hugging Face CEO's 2024 predictions, multi-modal AI on Vercel (11.27.23)

Friends, thought provoking pieces dominated the AI conversation this weekend and Monday. The US, UK, and other countries released a set of AI development guidelines they called "secure by design" but few in the AI community have discussed it with much interest yet.

As always, these are short summaries of the news most-discussed across the AI ecosystem, according to our weighted analysis of community engagement.

Today I'm starting a new section marked Early Stage for big news from small companies; we've got a good one of those in the software infrastructure market today!

And now here’s today’s news,

Marshall Kirkpatrick, Editor

Hugging Face CEO Forecasts Shifts in 2024 AI Landscape

First impacted: AI specialists, AI builders on Hugging Face
Time to impact: Medium to long

Hugging Face CEO Clement Delangue has predicted six key AI trends for 2024, and it was one of the most-discussed conversations in AI over the past few days: "- A hyped AI company will go bankrupt or get acquired for a ridiculously low price

- Open-source LLMs will reach the level of the best closed-source LLMs

- Big breakthroughs in AI for video, time-series, biology and chemistry

- We will talk much more about the cost (monetary and environmental) of AI

- A popular media will be mostly AI-generated

- 10 millions AI builders on Hugging Face leading to no increase of unemployment." That all seems like the direction things are going in; Delangue jokingly suggested in comments that it could be Hugging Face that goes bankrupt. [via @ClementDelangue] Share by email

Profile of Nvidia's CEO, Outpacing Walmart and ExxonMobil Combined

First impacted: CEOs of large US companies, AI developers
Time to impact: Medium to long

In May of this year, Nvidia's worth soared by around $200 billion in a day, making it the sixth most valuable company worldwide, exceeding the combined value of Walmart and ExxonMobil. This increase was sparked by the disclosure that ChatGPT was developed using an Nvidia supercomputer. Nvidia's CEO, Jensen Huang, is profiled in this New Yorker article. [How Jensen Huang’s Nvidia Is Powering the A.I. Revolution] Share by email

Simplified Transformer Block Boosts Deep Learning Efficiency

First impacted: AI researchers, Deep learning model developers
Time to impact: Medium

Researchers Bobby He and Thomas Hofmann of ETH Zurich, say they have simplified the standard transformer block used in deep learning models. The changes, based on signal propagation theory and empirical observations, have reportedly led to a 15% increase in training throughput and a 15% reduction in parameters, without impacting performance or training speed. This blew up over the weekend because Sebastian Raschka called it "easily among my favorite research papers that I've read this year." [Simplifying Transformer Blocks] Share by email

Google Colab Integrates Transformers by Default

First impacted: AI developers, Data scientists
Time to impact: Short

Google Colab, in collaboration with Hugging Face, has announced that its managed runtime images now come with transformers pre-installed. It's a significant improvement in ease of use, as developers can focus more on model development and less on environment management. [via @GoogleColab] Share by email

Observation: AI-Generated Content Dominating Search Results

First impacted: Internet users, Google search engine users
Time to impact: Short

AI-generated content is becoming more prevalent in search results, as noted by Ethan Mollick, who observed that the primary image for Hawaiian singer Israel Kamakawiwoʻole on Google is a Midjourney image from Reddit. Mollick, a Wharton professor now leading many conversations about AI use cases, said on X: "Seriously, don't trust anything you see online anymore. Faking stuff is trivial. You cannot tell the difference. There are no watermarks, and watermarks can be defeated easily. This genie is not going back in the bottle. One side effect from AI is that the corpus of human knowledge from mid-2023 on will have to be treated fundamentally differently than prior to 2023." [via @emollick] Share by email

XTX Markets Launches $10mn AI-Math Olympiad Prize For Shared Models

First impacted: AI developers, Mathematicians
Time to impact: Medium

XTX Markets has launched a $10mn fund, the AI Mathematical Olympiad Prize (AI-MO Prize), to encourage the development of AI models for mathematical reasoning. The company will award a $5mn grand prize to the first publicly-shared AI model that achieves a gold medal standard in the prestigious (founded in 1959) and challenging International Mathematical Olympiad and up to $5mn in progress prizes to models that hit key milestones. [AI-MO Prize] Share by email

Early Stage - Big News From Small Companies

Fal.ai Launches Integration with Next.js and Vercel

First impacted: Web Developers, Software Engineers
Time to impact: Short

Cloud inference startup Fal.ai has launched a new integration with Next.js and Vercel. This means developers using Vercel will be able to use models supported by Fal.ai to more easily add image generation (Stable Diffusion), speech transcription and translation (Whisper), and animations to their apps, including through a websocket-connected real time client. [Add fal.ai to your Next.js app] Share by email

That’s it! More AI news tomorrow.