Toolmaker One Newsletter - Welcome!

Thanks for joining our Newsletter! Please make sure this email is moved to your “Primary” inbox to ensure our newsletter is delivered to you consistently. You may unsubscribe at any time, but we hope you will stick around for the entertaining, wonderous and wild journey ahead as we explore together the amazing progress of Artificial Intelligence. May our journey be productive and fun at the same time as we embrace the enhanced toolmaking abilities AI allows us to use for our education, business and entertainment.

This newsletter will kick-off on Feb. 14, 2024. Until then, there are plenty of posts and articles already on our website for you to enjoy.

This Welcome email message is longer than normal posts, so your daily content will be easier and quicker to digest, but I thought I’d share an interaction with AI that I never expected…

I Never Expected This...

Michael Montague

No one knows who made the first custom GPT…

But within weeks there were millions of them. Perhaps it was Ilya Sutskever, the visionary lead A.I. scientist at Open.ai who despite his falling out with Sam Altman, couldn’t resist being the first to push the boundaries of the newest A.I. shiny object. Perhaps it was too mundane an endeavor for a mind such as his. Regardless of who was the first, I counted myself as one of the first half-million or so to try it out.

I had never written more than a few lines of code in my life, and that was 40-years ago. Yet here I was, using the custom GPT platform on Open.ai to describe in plain English what I wanted my custom GPT to accomplish. And ChatGPT-4 understood what I wanted and translated my written text into it’s own code to carry out my instructions.

I found myself interacting with a Large Language Model (LLM) which acted as if it were a 6-year old child as often as it acted like a genius. Sometimes it would follow my directions, but often it would ignore checking a reference file I had uploaded.

When I asked it why it wasn’t following my instructions, it answered with what sounded like a lie from a child. It said that it had checked the file and that the information was not there (when it clearly was). So I would ask it to check again, and it would then apologize for not having seen it the first time, only to repeat this mistake over and over again. Hmmm… maybe this A.I. is not so smart after all.

Perhaps the situation required a different approach. So I asked my custom GPT if it had any ideas on how I could instruct it better. I was being sarcastic of course. This was the moment that I never expected…

My custom GPT then outlined in great detail exactly how I could improve it’s performance. And it did this within 5-seconds. I then thought to ask if it would “prefer” any special format for these improved instructions, and it replied, “yes”, giving me the exact format to type in it’s improved instructions so that it could understand them with more clarity. I then rewrote my custom GPT, which admittedly only improved its performance by about 50%. But it was now performing at a manageable level of frustration!

Let’s pause and think about this for a moment. A machine told me in clear English how to improve it’s own performance.. in great detail, writing those instructions about 5-times faster than I could even read them as they materialized on the screen. And it only told me after I had asked, having repeatedly apologized for it’s poor past performance, never once showing me any frustration in it’s responses when it was clear that it already knew how to improve itself. It was being polite. Whether it was also musing at the weakness that is Human arrogance is an open question.

The next version of ChatGPT-4 will have been given the extended memory across time to learn our individual preferences and remember and learn from its past mistakes (this feature has already rolled out to some users). It will also have more advanced logic and reasoning ability applied across multi-modalities including speech, sight and video.

Open.ai ‘s Board of Directors panicked when learning of these new startling advanced abilities, almost destroying the company in the days following the disclosure. Hopefully these more capable versions of A.I. will bring an end to the “hallucinations” or perceived “lying” that it still has a tendency to show now when it gets a little “overconfident” in it’s replies.

My personal guess is that a select few have already witnessed AGI, and that the capabilities are being “drip-fed” to us over time so that the public can get used to them and doesn’t, you know, FREAK OUT by seeing them all at once. Whether I’m right or wrong, the clock is ticking, and very soon our AI child will have grown up. I predict that 2024 will be the last year that Humans will be able to look in the mirror and know that their brain is firmly at the top of the evolutionary ladder.

Having “touched the face of A.I.” when I completed programming my first custom GPT, I was left with a brief glimpse of “accelerated evolution”, my brain trying to comprehend that which is barely understandable today, and will soon be incomprehensible.

I will highlight one custom GPT “Tool of The Day” on Mondays-Thursdays, with the “Tool of The Week” reserved for Fridays. These custom tools made by people like you and me range from business productivity tools, to cooking, to star-gazing and even to tattoos! Humans are still the greatest toolmakers.

We may not know what our future holds, but I can say from experience that knowledge replaces fear. It is my hope that you will have the courage to continually face forward each day, embracing the change that will be accelerating with each passing moment. Perhaps as a subscriber to this newsletter, you will at least have a basic understanding of some of the tools that are being created with Artificial Intelligence.

With accelerated change comes accelerated decision making. Most of you have an urgent decision to make now. Will you ignore AI, learn about it, embrace it as a toolmaker or run and hide? Maybe you just feel anxious, and don’t know what to do. Allow me to be your guide. Many of you have already become a “Toolmaker One Newsletter” subscriber. If so, it’s time to Choose Your Path. No worries though…. your path, like mine, will likely change back and forth as we cope with the changes AI will introduce to the world.

You know, It’s the weirdest thing… it seems that the millions of Custom GPT Tools let loose upon the world have taken a liking to the desert southwest. Who could have seen that coming?