What you may not realize is that some form of artificial intelligence has been around as long as we’ve had computers. Alan Turing in 1950 proposed a “Turing Test” that could determine if a computer was intelligent or not, and ever since humans have attempted to create computer programs that were indistinguishable from human thought. We have had glimpses of AI in our lives for years, with digital assistants running our smart homes and helping to set timers while our hands are covered in dough.

Beyond simple prompts, we have had computers helping to identify things for us for many years also. I’m missing the details, but I remember hearing on a podcast about the rise of NVIDIA and AI about a couple of guys that were able to use a GPU for the first time to accelerate image processing that led to the start of CUDA in the early 2000’s. Once engineers were able to utilize the width of GPUs for machine learning, tasks that would have taken weeks started being done in hours, leading to today’s large language models (LLMs).

When we talk about AI now, we’re referring to generative AI, or Gen AI, which uses absolutely massive LLMs the size of which are hard to conceptualize. We’re talking about 10’s of billions of parameters that take massive super computers weeks to sift through. What comes from all of this processing is a model that can then be used to “understand” your speech and return answers or even operate things for you. Once the model is created, you can further train it with personal data to make it more useful. We call that inference, and it is why you now have things called Neural Processing Units showing up in PCs and phones. You have a huge batch of data, boil it down to a model, then sprinkle a little personal data on top and get a useful assistant.

Gen AI today is still very much so in its infancy, as is our understanding of what we can do with it. I was able to sit down with a Microsoft engineer earlier this year and asked him what people are doing with Copilot there. He mentioned using it to comb through the hundreds of emails he gets to find important information, but also being in a bit of a wait and see mode with AI. I’ve also been asking our clients if they’ve been interacting with AI, and it’s been a mixed bag. Some are using it to help change the tone of an email, some are using it to help find information, while some are not using it at all.

During this whole process of rebirth of Nordic, I’ve been using Copilot quite a bit. While writing our core story, I was able to get insights into how my story came across. I was able to take pages of writing and turn it into concise messaging. And all of the images for my blog posts have been created by giving Copilot a simple prompt: “give me a minimalistic thumbnail for this blog post.”

One thing is obvious, AI isn’t going away any time soon, and it’s just going to keep getting better and more useful. Probably the best thing that anyone can do now is just start using it a little bit every day. Just remember, it’s not perfect and can still make stuff up, so don’t take everything it tells you as the truth.

-Nate