Should we fear AI? After CEO of OpenAI...

The World’s Media has presented the CEO of OpenAI himself, stating, 'we should be afraid of ChatGPT'?

Why would he say this? Because we should be! This technology has changed the face of the world economy, as I have written in previous newsletters. Artificial Intelligence will change the economic structure of the world. The world will move a more in way back to the pre-Industrial Revolution days when trade and agrarian society were the main jobs for humanity. Since the day of Henry Ford and the Industrial Revolution, this has changed. So, the fear should be NOT that jobs will disappear *yet*, but the industry you work in could disappear. Other industries are 100% going to change.?

Two industries in which I have worked have already changed as a result of AI but will continue with the emergence of ChatGPT, LaMDA, Llama, and other competitors. Those are Cybersecurity and Computer Programming. The emergence of AI in Cybersecurity has not come with ChatGPT, but with the ChatGPT API and other APIs, custom-created policy, and AI-based systems that learn, adapt, and secure the system using data rather than seeing the notification, and researching it, finding it to be a risk, and then begin mitigation. AI allows for immediate recognition with instantaneous mitigation. This is a simplification of the system and the method of operation, but just look at tools like @Darktrace. The only difference is that this is built on a private AI.

Computer Programming is going to change as technology grows, and the code written works instantly without having to worry about it being old code generated from incorrect solutions. Computer programming will go from writing the application from head to toe to writing sentences of the various pieces of the application that ChatGPT generates, and moving these pieces together and creating what took years in days. Technology is changing the world and will continue to as robotics, machine learning, quantum computers, and other technologies continue to evolve.?

But AI is still in a very early phase, and we have yet to see public releases of Quantum AI, the Quantum Cloud, and Clustered or Neural AI. We have yet to see many technological advancements, potentially existing or in labs somewhere. So, humanity needs to present the questions a little earlier. The fear of AI was discussed in the Obama years, but the actual impact and thought. With a heavy focus on what to do regarding the technology that has been in the making for years, it has upon its 'public' release.?

While referring to hardware, Moore’s Law ceases to exist GPT-3 was released in beta in June 2020, and the API in September 2020. This growth is what I am referring to when saying Moore's Law is being eliminated with this technology, and like I said, Moore's Law relates to Hardware like the Quantum Computer. Interestingly, this was not even of the predicted technologies to emerge in 2022. But not only was it released, but it’s publicly released and for sale to private parties. The Quantum discussion is a side note but one that relates to the main topic of ChatGPT in the sense that Quantum AI will emerge even sooner now that private industry can obtain even smaller qubit quantum machines.?

Back to ChatGPT and its evolution of it through the years:

• GPT-1: 117 million parameters
• GPT-2: 1.5 billion parameters
• GPT-3: 175 billion parameters
• GPT-3.5: 6 billion parameters

Evolution from 117 million parameters to 1 trillion, ChatGPT has increased its capabilities with this expansion to include something that previously did not exist. OpenAI’s release came with a new feature the multi-modal input demonstrated something, be it power, drive to stay on top or fear of the competitor. Still, an over five-fold increase in the parameters is incredible, as first, it shows the parameters were already waiting. Still, also it shows the algorithmic changes needed to compare the hundreds of billions to generate the answer to over a trillion. But why stop there? I foresee 10s of trillions of parameters in the next iteration. Including more and more detail for sound, image, video, and instantly transmitted. I would love to see the actual scalability of ChatGPT. I can only assume that it is running Kubernetes and, of course, being Microsoft on an Azure VPC Cluster, spinning up a new VM, most likely for every user, and a new instance of the application upon every new question. What I wrote above is an assumption but with the resources and companies that now operate and maintain GPT.

This week, we have witnessed the resources being this week with iteration version versions 3.5 and 4.0 released within days. What's incredible about this is the expansion from 175,000,000,000 to 1,000,000,000,000 parameters.? That is a nearly five-fold increase in the parameters in mere days.

The partnership with Nvidia comes in with this expansion because of the extreme need for GPU capabilities, mathematical processing, and yet to start using a quantum computer. The architecture is moving towards clustered GPUs used in the computational aspect of Deep Learning. Taking those trillion parameters and calculating the NLP, Natural Language Processing, and comparing it to the parameters repeatedly eliminating those in no way connected to the question. This is shortened for explaining sake. The next iteration has the parameters not iteratively calculated down to a series of parameters mathematically closest to the entropy of 0. These parameters are introduced as a result set to an NLP to produce the answer.

?When can you ever say something is good or bad? You will be until they’re blue in the face, but the genie is out of the bottle in the case of AI. Secondly, the problem is that there is neither good nor bad.? It is based on algorithms written by a man. Programming by man inherently depends on the programmer, calculations, and the data ingested. From there, it depends on how AI is used. Any tool, from the hammer to AI, can be used for good or bad.? Which brings do we police it, and what do we do with its capacity??