Google Gemini Drops, Meta Image Maker, Musk in Denial

First, a Daily Meditation

Today, the first thing I saw on LinkedIn was a post about Jane Goodall wishing she could erase most of humanity. A couple of days earlier, I read about Larry Page and Elon Musk falling out because the former called the latter a “specist” for preferring human life over machine life. What is wrong with us? I know some people are annoying, but hating all of humanity? Can’t we all just get along? I believe in AI, but sometimes I get down as I realize that there might not be enough natural intelligence for it to imitate. 

Speaking of which, every day, I see both wonderful and absurd chatter on social media regarding AI. One angry post I saw claimed AI is a bubble in the mold of blockchain. I disagree. I don’t think blockchain was a bubble, but that’s for another newsletter. I am convinced that AI is different. Anyone who has used it knows it is a game changer. No matter what your profession, you will find a use for it, and it is going to absolutely transform so many industries.

Ok, enough ranting. Here is today’s digest!

Joaquin

Google Surprise Drops Gemini

TLDR:

• Google unexpectedly releases Gemini

• Gemini is a powerful multimodal LLM

• Aims to catch up with ChatGPT

• Google feeling the heat, well behind in AI

Only a couple of days after announcing Gemini would not come out until 2024 (which we announced here a couple of days ago), Google decided to make us look bad by releasing it today, December 6, 2023. They claim that this LLM beats GPT-4 in 30 out of 32 benchmarks ( perhaps one of them being unexpectedness?). 

Gemini is multi-modal (in lay terms, you can give it audio, video, text, and images to interact with). It is available in Ultra, Pro, and Nano versions, with the bigger ones requiring more resources and the smallest version capable of running on the Pixel 8 Pro phone. You can try it out for text-based prompts within Bard, starting now. 

Google, or Alphabet, if you are so inclined, has famously been behind its competitors in the AI wars. This has led many to speculate that the tech giant’s entire business model could be under threat, as many users are turning to alternatives such as GPTs and TikTok instead of the good old Google search. Only will tell if Gemini will be the first step in successfully revamping Google’s product ecosystem. 

Meta Introduces the Imagine Image Generator

TLDR:

• Meta Launches Imagine, a new standalone generative AI 

• Image generation in natural language

• Free to use (US users)

• Includes invisible watermarks that identify AI-generated images

On Wednesday, Meta announced yet another AI product: Imagine. Imagine is an image generator that can create images based on text prompts. Meta has been stepping up its AI game, introducing the widely-acclaimed open-source Llama earlier this year. They have also shipped products for music generation and translation. Imagine steps into a crowded field with several dominant players, such as DALL-E, Midjourney, and Stable Diffusion. An early Gizmodo test resulted in a rather negative review. We tried to review it, but our attempts resulted in a “Sorry, something went wrong” error. Sigh.

Musk Denies Reportedly Seeking $1 Billion for xAI

• Elon Musk working on ChatGPT rival, Grok

• Previous reports claimed xAI seeking $1 billion

• Musk denied the reports a day later

Lately, the world of AI has truly embraced a tabloid-worthy attitude. Dramas, intrigue, conflict between humanists and AI-ists, and more. True to fashion, Elon Musk has jumped into the fray. On Tuesday, several media outlets reported that Musk was seeking $1 billion in funding for xAI and its star LLM, Grok. But the next day, CoinDesk reported that Musk had tweeted X’d posted on X that xAI was not seeking funding. 

It is unclear what Musk means by this. Does his denial mean that he is going to provide the funding? Or could it be that xAI already has the money? It could also be due to some intrigue related to an SEC filing that led to a spike in dogecoin values. Whatever the case, many (ourselves included) ​​are eager to test out Grok, which at least promises to be sassier and less politically correct than ChatGPT. You might remember that the author of this newsletter had some absurd censorship issues with ChatGPT/Dall-E 3 a while back.