First came the AI Big Bang; now comes the aftershocks and tremors... TL;DR and podcast 10Dec23

Last update: Sunday 12/10/23 

Welcome to our 12Dec23 TL;DR + podcast about the past week's top 3 AI stories on our "Useful AI News" page 1) The biggest aftershock: Google's Gemini, (2) Tremor: Meta and IBM lead an alliance, and (3) Another tremor: Apple’s new framework 

Demis Hassabis, PhD -- CEO, Google DeepMind

Click the "start" button in the audio control (below) to hear the podcast ...  If audio fails to start or gets stuck, try reloading the page. 

TL;DR link ➡ HERE

Background ...

Readers of this blog may recall that "The AI Big Bang" was the name we gave to the long, tumultuous weekend that began on Friday evening 11/17/23 when the OpenAI board abruptly notified its CEO Sam Altman that he was fired. It ended on Tuesday evening 11/21/23, when Altman returned to OpenAI as CEO

So who won this brief high tech battle? Microsoft and its CEO Satya Nadella won handily. Microsoft's stock hit an all-time high after Altman's return was announced. 

Microsofts supremacy was quickly challenged by its leading Big Tech AI competitors -- Amazon, Google, Facebook, and Apple.  Their complex challenges were under preparation long before the Big Bang, but the competitors were pressured to release their challenges before they were fully completed, presumably to provide their investors with timely reassurance that their AI programs were still viable. 

Most of the challenges were incremental tremors. Only one challenge was an aftershock whose innovative magnitude was comparable to Microsoft's Big Bang

To be specific, Amazon's announcements at its "re:Invent" conference last week was a tremor; and as our reader will soon see, the challenges from Facebook and Apple this week were also tremors. Only Google's challenge was an innovative aftershock.

1) Google announced "Gemini"... its new LLM

Gemini comes in three versions: Gemini Nano, Gemini Pro, and Gemini Ultra

• Nano, the lightest version, will eventually run on all Android devices, but currently only runs on Google's Pixel smartphones
  
  
• Pro is more powerful than Nano, and currently runs Bard on all devices. Bard can be accessed at bard.google.com
  -- Pro also runs Google's NotebookLM application, an innovative AI driven app. NotebookLM can be accessed at notebooklm.google.com
  -- Pro will be available to developers and enterprise customers in mid-December via Google AI Studio
  
  
• Ultra is the most powerful version, but Google has not yet announced a release date
  
  
• At present Nano and Pro accept text input, then deliver text output, but all versions will eventually be multimodal, accepting input as text, images, video, audio, and code.
  
  
• All models were trained on Google's own chips (Tensor Processing Units) and will run in data centers on those chips
  
  
• According to The Verge, "Hassabis says. 'There’s still things like action, and touch — more like robotics-type things.' Over time, he says, Gemini will get more senses, become more aware, and become more accurate and grounded in the process. 'These models just sort of understand better about the world around them.' These models still hallucinate, of course, and they still have biases and other problems. But the more they know, Hassabis says, the better they’ll get."
  
  
• Benchmarks
  Google's official announcement presented the results of 18 benchmarks tests that compared Gemini Ultra with GPT-4. Gemini performed better than GPT-4 by small margins on all but one test.
  
  
• Stock market ... the most important benchmark???
  "Alphabet soars as Wall Street cheers arrival of AI model Gemini", Aditya Soni, Reuters, 12/7/23
  
  

2) Meta, IBM, and others announce the "AI alliance"

According to the mission statement of the AI Alliance on its website: 

"The AI Alliance is focused on accelerating and disseminating open innovation across the AI technology landscape to improve foundational capabilities, safety, security and trust in AI, and to responsibly maximize benefits to people and society everywhere. 

The AI Alliance brings together a critical mass of compute, data, tools, and talent to accelerate open innovation in AI."

There are more than fifty initial members in the alliance:  corporations, universities, non-profits, and government agencies ... in the U.S., Germany, Italy, Japan, Israel, India, UAE, UK, Australia 

Its mission sounds uncomfortably like the mission of OpenAI when it was founded back in 2014. But as we all know, OpenAI soon became  Microsoft's well-funded R&D subsidiary. So will such a large alliance prove to be too unwieldy to be effective? If so, might it be then captured by its corporate sponsors? 

3) Apple announced a machine learnning framework running on its own chips

Whereas Microsoft and Google are producing AI tools that can be used today by non-techs, Apple's MLX machine learning framework is a tool that machine learning techs can use to generate models that will provide AI services to non-techs tomorrow ... but probably no time soon.

B. Top 3 stories in past week ...

1. Google
   "Google DeepMind's Demis Hassabis Says Gemini Is a New Breed of AI", Will Knight, Wired, 12/6/23 *** 
   -- This story also covered by The Verge, MIT Tech Review, NY Tmes, Ars Technica ... and Google DeepMind
   
   -- Fake demos in Google's announcements: "Google just launched a new AI and has already admitted at least one demo wasn’t real", Emilia David, The Verge, 12/7/23 ... Also reported by TechCrunch, Gizmodo, BBC
   
   -- Users access Gemini Pro via Bard: "How to Use Google’s Gemini AI Right Now in Its Bard Chatbot", Reece Rogers, Wired, 12/6/23
   
   -- Google's new Gemini powered notebook: "Google’s AI note-taking app is now available to users in the US", Emma Roth, The Verge, 12/8/23 ... Also reported by Engadget, Gizmodo, 
   
   -- Announcements boosted Google/Alphabet's stock: "Alphabet soars as Wall Street cheers arrival of AI model Gemini", Aditya Soni, Reuters, 12/7/23
   
   
2. LLM News
   "Meta and IBM Launch AI Alliance, Promise to Be ‘Open’", Maxwell Zeff, Gizmodo, 12/6/23 *** 
   -- This story also covered by Computerworld, TechCrunch, Fortune, Ars Technica, Bloomberg, 
   
   
3. Non-LLM News
   "Apple Scurries Onto the AI Stage With Too Little, Too Late", Maxwell Zeff, Gizmodo, 12/7/23 ... the MLX machine learning framework that runs on Apple's own chips  *** 
   -- This story also covered by The Verge, Computerworld

C. Who is Demis Hassabis, CEO of Google DeepMind? ... Two interviews

• "Inside Google’s big AI shuffle — and how it plans to stay competitive, with Google DeepMind CEO Demis Hassabis", Nilay Patel, The Verge, 7/10/23 ... audio and transcript of interview 
  
  
• "A.I. Could Solve Some of Humanity's Hardest Problems. It Already Has.". Guest = Demis Hassabis, The Ezra Klein Show (podcast with transcript), 7/11/23 

D. Four breakthroughs in science/technology using DeepMind's AI/Neural Networks

• "One of the Biggest Problems in Biology Has Finally Been Solved",  Tanya Lewis, Scientific American, 10/31/22 ... protein foldimg problem ...DeepMind
  -- This story also covered by VentureBeat,  CNET, Science, 
  
  
• "Courtesy of AI: Weather forecasts for the hour, the week and the century", Devin Coldewey, TechCrunch, 11/14/23 ... DeepMind
  -- This story also covered by Gizmodo, Financial Times, MIT Tech Review, Ars Technica, Wired, ZDNet, 
  -- These articles describe research published in Nature and Science
  
  
• "AI Tool Pinpoints Genetic Mutations That Cause Disease", Even Callaway & Nature Magazine, Scientific American, 11/21/23 ... DeepMind
  -- This story also covered by Wall Street Journal, Wired, 
  
  
• "Google’s DeepMind finds 2.2M crystal structures in materials science win", Michael Peel, Ars Technica, 11/29/23 ... "The trove of theoretically stable but experimentally unrealized combinations identified using an AI tool known as GNoME is more than 45 times larger than the number of such substances unearthed in the history of science, according to a paper published in Nature on Wednesday."
  -- This story also covered by Wired,  VentureBeat, Nature, Science,

D. Links to ChatGPT's responses to five basic AI questions:  

ChatGPT's answer to each question provides the framework for its response to the next question. So most readers should start with Question 1.

• Question 1: What is Artificial Intelligence?
  
  

• Question 2: What is Machine Learning?

   

• Question 3: What are Neural Networks? 

  
  

• Question 4: What are GPT models? 
  
  

• Question 5: Have models other than GPT made substantial progress towards AGI? ... ChatGPT's response limited to its knowledge of models in 2022.