- TOMORROW'S APE
- Posts
- Open Small Cheap vs Closed Huge Expensive
Open Small Cheap vs Closed Huge Expensive
The battle for the model Model š¤
3Images in this Newsletter made with ideogram.aiĀ
Hereās what weāve got this week:
The latest Fallen Fruit
Open Small Cheap v Closed Huge Expensive
Wheels in Motion: AI Video beyond 48 frames
Smart Takes: Swinging from the High Branches
Forecasting the Future with Metaculas
Awe Gazing @ latest Tech
Retro Movie Rewind
FALLEN FRUIT |  |
 ChatGPT can now hear, see and talk! This is big news off the back of the release of DALL-E3. ChatGPT has been amped up with Multimodal capabilities which has everyone excited. It even looks capable of coding simple web pages off of a sketch.  Amazon invest $4B into AnthropicAI We had Microsoft + OpenAI Google + Deepmind Now this! The big companies are not wanting to be left behind without high level AI integration. It does leave everyone wondering where Apple is at. šÆš¤Tesla Optimus Optimus does Yoga and shows just how elegant the robots moves has become in what it is a relatively short development time.  šHugging Face Hugging Face have released a guide for non-coders on how to train LLaMA2 chatbot. Pretty awesome to see this coming out and would be great if this is a new trend to help businesses with AI adoption.  Meta have released a demo of their latest Smart glasses with Ray-Ban. |  We now see the vision of Photorealistic Avatars in the Metaverse in this Lex Friedman podcast with Mark Zuckerberg. It does look like this will be the Video Calls of the future.  Itās been reported that OpenAI & Jony Ives plan to make an AI phone with $1B backing from Softbank CEO Masayoshi Son. Jony Ives was lead designer at Apple and a key part of the team that brought you the original iPhone. If your interested in a companies beginning the process of building a Nuclear Fusion Power Station. This video features Robert Mumgaard CEO of Commonwealth Fusion Systems & David Kirtley CEO of Helion Energy  Itās pretty cheesy bro talk, but this is the world we are living in now. |
Open Small Cheap v Closed Huge Expensive | A Short Summary of the Open v Closed Source Enterprise Landscape |
There has been speculation this week about what the best applications and forms in which AI serves businesses.
The next frontier, business 2 business has been slowly introduced from Large Closed Model companies, like OpenAI with Enterprise and Amazon with Bedrock.
The Hugging Face Community, the worlds central repository of Smaller Open Source Models have also been busy developing options too.
The arguments:
Large Closed Models like ChatGPT are highly capable, but not easily integrated or manipulated for each individual companies needs. Along with this, anyone who does try to bootstrap an API from it are at the mercy of the companies like Open AI should they change in the future. | Smaller Open Source Models offer more āownershipā over the model and therefore more control and ultimately more privacy guarantees as well. They are also dynamic, with a community able to share their knowledge and iterate on the fly. |
This will be a watch and see scenario as businesses attempt to take advantage of AI to gain a competitive edge.
WHEELS IN MOTION | VIDEO BEYOND 48FRAMES |
For all of history this āpictureā guy has stood stillš§š½āāļø
Now with AI you can make him do this šš½āāļø or even thisšŗš½
Ah, I hear you say. We already invented motion capture over a hundred years ago! Not to mention animation!
Yes, you would be correct, but now the ācaptureā part is gone and the need to for exhaustive animation processes too. Yay!
You only need one still image and text for instructions.
Itās a pretty big deal. No more building sets, hoards of crew members, extras, lighting rigs, catering vans, expensive camera equipment or an army of animators for that matter.
Anyone with an idea for a film can in theory, have it realised without any of these.
But, at this stage what we are seeing are short films with short edits.
How are they made?
In short
A series of still images that serve as final edit quality storyboards (as opposed to sketches in traditional film making) are created by a Diffusion Model like MidJourney.
Then each shot is turned into a motion by AI.
Creators can really hone their shots, generating very high quality cinematic compositions first before animating them.
The resulting animations are limited by the abilities of the AIās to predict correct motion of objects, characters and backgrounds. This is one of the reasons for the short edits.
Whoās got the best platform to create AI films? Here are a few that weāve seen promising results from so far:
With the latest release of DALL-E and itās integration with ChatGPT, We expect to see some larger AI companies bring their own video generators given the potential.
We already have the capabilitites to augment voice and mouth movements, but once the models are able to more seamlessly replicate human body movement and expression, a new horizon may be born.
Check out Curious Refuge started by Caleb & Shelby Ward, creators of some of the first AI films to hit the publicās imagination. They have called Curious Refuge the āFirst Home of AI Filmakingā.
You can also check out Calebās talk about AI Filmaking at the recent All-In Summit here.
SEEING INTO THE FUTUREš®ā¦ PROBABILISTICALLY š¤
Metaculus is forecasting and modelling service for forecasters, decision makers and the public.
Why do we like it? They specialise in AI, and cos itās fun to think about things like, when will we reach AGI and when will the singularity occur?
They have aggregate forecasts on a range of AI related questions.
Here are just a few as they stand at the time of writing this:
  |   |
Itās free to sign up & you can have a go at predicting yourself!
SWINGING FROM THE HIGH BRANCHES |  |
 History vs FutureThe two intellectual giants disagree respectfully about the potential for AI catastrophe.  LLMās will be the next OS The guru of AI Andrej Karpathy has laid out a pretty insightful projection about LLMās and their future. āTLDR looking at LLMs as chatbots is the same as looking at early computers as calculators. We're seeing an emergence of a whole new computing paradigm, and it is very early.āItās worth a full read though! |  Open Source is winning?! Clement Delangue CEO of Hugging Face has been busy on Twitter espousing the advantages of Open Source over closed. āOpen science and open-source is the foundation of all AI. The rising tide that lifts all boats!āāContrary to popular belief, a company (or a country) sharing their research, models and datasets publicly in open-source makes them MORE competitive, not LESS, even more so in AI.ā  Had to include this great summary of Crypto. The general populous doesnāt understands it, so this is a pretty good summation by Peter Van Valkenburgh, Research Director at Coincenter. Also a good summary in there about Web 1, 2 and 3. |
AWE GAZING @TECH
Heygen released a video by popular demand of Ronaldo Speaking English An amazing tech that will break down boundaries. |  |
 | Starlink SatellitesĀ like fireflies in the sky on this animation of all the current starling satellite flight paths. Clearly Starlink and Space X have been busy. |
Drawing a website - Sketch and your done! ChatGPT4 can kinda do the job. Looking forward to trying this first hand. |  |
A young man finds a back door into a military central computer in which reality is confused with game-playing, possibly starting World War III.
It features an AI one year before Terminator came out. The main character hacks into to the militaryās ānon-humanā control program āJoshuaā, communicating with it like a Chatbot essentially. It has some poignant moments that ring true today in regards to an Artificial Intelligent Agent at the helm of technology that could destroy us.
What did you think about this weeks Newsletter?
Did we nail it? Can we do better? Want to know more? Hit reply and let us know!
If you like our content and think it might be worth a shout out, please do!
Right click the below button and share the link.