I’ve been a programmer for 5+ years. Not much by industry standards, but enough by my measure to consider myself in a rhythm when it comes to my job. I love doing it. It was my passion and I’m grateful I was able to make it a career. It’s something experienced by so few around me that I’ve always felt fortunate. I interacted with ChatGPT for the first time though, and I felt like my luck had run out.
My first reaction to ChatGPT-3 was a sense of wonder. I saw it generate code with ease to display a ball bouncing around the screen. It was crude and a little ugly, but it did what would have definitely taken me more than a few minutes to replicate. I felt like I was John Henry, but before I could even put the hammer in my hand, this AI had finished the race. Speed will not be a contest in the coming era of AI when it comes to building.
But then what does that leave us with? Quality then must be our last human-influenced barrier. I would say not. Just with this simple test, an AI remembers to write comments and follow core coding principles I would forget on a good day. End-user results can only get better with indistinguishable differences to front-end users, so quality will improve to be indistinguishable. Even with writing articles, it is getting harder to differentiate the human from the machine. So then are we doomed? Is the party over on the careers and businesses we’ve built? For the short time we’ve had to grapple with these questions in an Open AI world I’ve thought it was.
My solution and research have brought me to launch Fine-tune Sandbox. This perceived threat will displace jobs and professions, make no mistake about it! The ones that survive will be the ones that understand how to harness AI and make it their own. Those with originality and style will be those that thrive at the onset of this digital wave. The architects, inventors, and producers were always what we were meant to be. With the heavy lifting of coding, coloring, and shading done for us, we need only be able to imbue our mark within it. A mark and style no machination could come up with alone.
There is a process to train an AI model to more closely follow the responses we seek. That is what we call fine-tuning. It is a data model we create to give more context to the technology about the content we want to be created. It can have immeasurable value beyond just using Chapt-GPT alone. Take, for example, an article writer that likes to use a witty voice with cynical undertones in many of his pieces. If I were to right now give Chat-GPT a prompt that asked it to write an article with those qualities, it would do it. It might even have hints of sounding like a real-life writer, but not close enough. Now say, we took 100’s and 1,000’s of articles by this creator, and we fed them to Chat-GPT by this process of ‘fine-tuning’. Rather than give a prompt asking for a piece about ‘x’ topic with those characteristic undertones, I would only ask Chat-GPT to simply write a topic about ‘x’, and it would do so more closely matching the style of our writer. It might not be a perfect match, but the writer would be along guiding the AI closer and closer with each fine-tuning. Eventually, he would need only need create the topic and proofread the sentence structure to generate articles with his signature wit. Our core belief, every professional and service will need to preserve their design within these models that will enable them to build with never before seen speed and quality. Fine-tune sandbox is dedicated to creating these models alongside users with our software, as well as giving professional guidance by means of trained prompt engineers. Even if you don’t have the specifics of how to introduce this new technology into your process, we can help get you started in seeing where to begin. Whatever level of detail and monitoring your model needs, we have the means to provide it for you. Anything less than the best will not meet future standards of speed, quality, and style needed to survive the technology wave.
