Tuesday, September 10, 2024

ChatBots Are For Children: "What’s Ahead for OpenAI? Project Strawberry, Orion, and GPT Next"

Following on September 8's "ChatBots Are Not The Be-All And End-All Of Artificial Intelligence":

Far from it.
And all the focus on ChatBots and LLMs are more than just a distraction, they are a perverse representation of what AI is doing and will do and could potentially cost you money or opportunity or both....

For reference here is the roadmap OpenAI is pitching, via Bloomberg, July 11:

OpenAI Scale Ranks Progress Toward ‘Human-Level’ Problem Solving

....MUCH MORE

And from Decrypt, September 3:

What we know about the secretive AI projects pushing the limits of what OpenAI can do. 

OpenAI is on the cusp of releasing two groundbreaking models that could redefine the landscape of machine learning. Codenamed Strawberry and Orion, these projects aim to push AI capabilities beyond current limits—particularly in reasoning, problem-solving, and language processing, taking us one step closer to artificial general intelligence (AGI).

Strawberry, previously known as Q* or Q-Star, seems to be more than just a chatbot; it focuses on showing a significant leap in AI reasoning abilities. Sources familiar with the project have told different media outlets like Reuters or The Information that it has demonstrated remarkable proficiency in solving complex mathematical problems and enhancing logical analysis.

Orion, meanwhile, is positioned as OpenAI’s next flagship language model, potentially succeeding GPT-4. It's designed to outperform its predecessor in language understanding and generation, with the added ability to handle multimodal inputs, including text, images, and videos.

Both projects have garnered attention from U.S. national security officials, underscoring their potential strategic importance. This development comes as OpenAI continues to raise capital despite substantial revenue growth, likely due to the high costs associated with developing and training these advanced models.

Strawberry and reasoning power
Despite an unending flurry of speculation online, OpenAI has not said anything officially about Project Strawberry. Purported leaks, however, gravitate toward its capabilities for sophisticated reasoning.

Unlike traditional models that provide rapid responses, Strawberry is said to employ what researchers call "System 2 thinking," able to take time to deliberate and reason through problems, rather than predicting longer sets of tokens to complete its responses. This approach has yielded impressive results, with the model scoring over 90 percent on the MATH benchmark—a collection of advanced mathematical problems—according to Reuters.

Another key innovation anticipated from Strawberry is its ability to generate high-quality synthetic training data. This addresses a critical challenge across AI development: the scarcity of diverse, high-quality data for training models. If true, Strawberry not only enhances its own capabilities, but also paves the way for more advanced models like Orion.

Considering the huge amounts of data already scraped by OpenAI, and the privacy movement that is now very present among users unwilling to give their data to AI trainers, this feature may play an important role in the quality of future AI models—just like some users today train their own custom models using images generated by Stable Diffusion....

....MUCH MORE

Because the company is making up the roadmap and the nomenclature on the fly we are seeing some confusion in media perceptions and explanations. One recent hullabaloo was the leaked $2000 per month proposed pricing. Another is "GPT Next" Here corrected via DataConomy, September 6:

“GPT Next” isn’t OpenAI’s next big thing, yet it will be still powerful
OpenAI has clarified that "GPT Next" is not an actual product 
Regarding the pricing for the next product, understanding advantage flywheels and hyper-Pareto distribution of profits is crucial to avoid sounding like a complete moron when I speak about the business end of this stuff with people much smarter than myself:

In Nvidia's World, If You (and your company) Don't Have Money You Will Not Be Able To Compete (NVDA)

The advantage flywheels keep spinning and reinforcing each other to the point that the Pareto distribution of profits - 20% of companies reap 80% of the profits - is becoming Super-Pareto where 5% of the companies reap 95% of the profits and is approaching Hyper-Pareto at maybe 2% of companies reaping 98% of profits.

It all comes down to having the resources to keep up. 

I watched Mr. Huang give the keynote and it's all a bit much to digest before firing out comments that would make any sense at all so here are some of today's headlines to give a taste of what the intro paragraph is based on.