By: Adam Sharp
Editor, The HIVE Newsletter
Over the past few weeks I’ve been exploring ChatGPT, a new AI assistant created by OpenAI (co-founded by Elon Musk and Sam Altman, among others).
ChatGPT can create high quality content of almost any type. It can do homework, write a business plan, or simulate a paralegal.
ChatGPT can even write code. Here’s what that looks like.
That’s just a simple example. This AI is capable of writing far more advanced code. You can have it explain any part you’re confused by, and it can translate what the code is doing into easy-to-understand text.
In its first five days of public access, ChatGPT attracted more than 1 million users. Here are a few comparisons showing how long it took big tech companies to get to that user base.
Using this tool offers a glimpse into our AI-powered future. ChatGPT is just the first powerful AI using large language models (LLMs) to go public. I believe in the near future we’ll all have our own personal AI assistants. And they won’t be “dumb”, like Siri or Alexa are today.
These new assistants will be like ChatGPT, and they’ll be incredibly useful for all sorts of personal and professional tasks.
Competing AIs using this same tech will be released soon. And the more these models are used by people, the better they will become over time.
I believe this tech is going to change the world quite a bit.
How disruptive could it be? Well, the creator of Gmail, Paul Buchheit, says “total disruption” for Google may only be a year out. Here’s what he recently tweeted.
Google may be only a year or two away from total disruption. AI will eliminate the Search Engine Result Page, which is where they make most of their money.
Even if they catch up on AI, they can’t fully deploy it without destroying the most valuable part of their business!
This new type of AI is going to be an extremely disruptive force. I think it will change the way we learn, teach, research, treat disease, and conduct business.
AI = Powered by GPUs
These remarkable advances in AI are powered by supercomputers with thousands of GPUs (graphics processing units).
It takes a LOT of GPU power to run ChatGPT. For instance, OpenAI CEO Sam Altman recently tweeted that the computation cost (per single chat) is in the “single-digit cents”. That’s a high cost compared to a typical search engine. If AI applications are to eventually reach the scale of Google, a ton of GPU brute-strength will be required.
And the new breed of AI image generators such as OpenAI’s DALL-E and the open-source Stable Diffusion may also add to demand for cloud-based GPUs for AI applications.
HIVE Blockchain happens to specialize in GPUs, because from 2017 to late 2022, the company used about 38,000 of them to mine Ethereum.
When Ethereum recently switched to a proof-of-stake consensus system, the age of ETH mining ended. Since then, HIVE has been exploring ways to leverage its fleet of GPUs, and its expertise in running data centers.
Let’s ask ChatGPT if HIVE’s high-end NVIDIA A40 GPUs are suitable for AI applications.
There’s something there worth exploring. This new breed of AI is one of the most promising potential uses for GPUs. We will keep readers posted as we continue to delve into this exciting area.
Cheers,
Adam Sharp
Editor, The HIVE Newsletter