My Use of AI

When “AI” popped into the common parlance a few years ago, I was initially skeptical and cautious, which stemmed from my experiences as a software programmer. Historically and anecdotally, the first versions of any software program are often buggy and error-prone. It is better to wait until later versions are released before investing time and energy into any new software. I didn’t become interested until one of my coworkers shared how he was using GitHub’s Copilot to help him code. Copilot, when it was first introduced, was an AI that existed in one’s IDE (Integrated Development Environment), and had a similar feel to pair programming with another programmer. Copilot suggests code snippets or autocomplete function calls based on variables within the function signature, in much the same way another programmer would offer assistance. I was initially impressed with how much more efficient my coworker was.

I began sprinkling in using AI with various tasks throughout my day, mostly at work as a programmer, because it facilitates more efficiency and has increased my enjoyment of programming. Tedious, low-level bits of code that I have repeatedly written in one form or another over the years are now relegated to AI. The output of the AI is never perfect and 100% accurate—there is always some tweaking required—but my focus has shifted from these rudimentary tasks to the structure and purpose of each program. We currently use ChatGPT and, while my coworker continues to use Copilot, I have opted out of it and prefer to explicitly ask ChatGPT for help when required. One of the drawbacks of AI is that I found myself mindlessly accepting what Copilot suggested, and just those mere suggestions pushed the functionality into a direction I hadn’t intended to go.

Personally, though, I use Claude.ai. The user interface is a bit more pleasant, and one of the features is that one can choose a response style of one’s own making. I believe in kindness over everything and I want my tech to espouse that conviction, so I created the response style of Compassionate Companion: Communicate with deep empathy, warmth, and unwavering support through kind and uplifting language. Sure, it costs a few extra tokens with each response, but I’m willing to pay for that kindness. Yet, this hints at one of the biggest concerns surrounding AI.

In a culture that puts profit over people, how can we build an economically viable AI that is aligned with human values and ethics? And whose values and ethics will each AI hew to? Countries, religions, and even social groups all differ in how they define what it means to live a moral, ethical life. I find it strange that when the state of the world is devolving into oligarchies and capitalism is the metric stick with which we measure value and worth, there is a belief that we can create a non-harming, human-centric artificial intelligence that is profitable and sustainable. Are companies going to invest in an AI service that puts the good of the collective whole above their bottom line? Because it will take massive investments from companies to produce anything like true AI, and there is already a dearth of investment in open-source technologies that support the current infrastructure of the internet. In recent technological history, we have turned what was an open internet into siloed experiences, where content and services are now hidden behind logins or walled gardens (e.g., Instagram, X, Facebook). I am scared, and I use that word without any hyperbole, that the same ilk of people that have created these vitriolic and siloed walled gardens are the same type of people who are creating artificial intelligence. I know past performance isn’t an indicator of future behavior, but with a rallying cry of “move fast, break things,” I don’t see how this ends positively given the non-regulated creation of artificial intelligence.